Prev Page 1 Next

T HE N ETWORK P RESS
ENCYCLOPEDIA
OF NETWORKING
W ERNER F EIBEL
NOW IMPROVED -THE MOST COMPREHENSIVE
COMPENDIUM OF NETWORKING CONCEPTS ,
ISSUES ,
AND TERMS
COVERS NET WARE 4.1, W INDOWS 95,
WINDOWS NT S ERVER 3.51, U NIX WARE 2,
AND OS/2 W ARP CONNECT
FULL TEXT ON CD-ROM FOR QUICK
ELECTRONIC REFERENCE
S ECOND E DITION


Prev Page 2 Next



Prev Page 3 Next




The Encyclopedia
of Networking




Prev Page 4 Next



Prev Page 5 Next




The Encyclopedia
of Networking
Second Edition
The First Edition of this
book was published under
the title Novell's Complete
Encyclopedia of Networking

Werner Feibel


San Francisco I Paris I Dsseldorf I Soest


Prev Page 6 Next


Acquisitions Editor: Kristine Plachy
Developmental Editor: Guy Hart-Davis
Editors: Kristen Vanberg-Wolff and Maureen Adams
Technical Editor: Mary Madden
Book Designer: Seventeenth Street Studios
Technical Illustrators: Cuong Le, Heather Lewis, and Alan Smith
Desktop Publisher: London Road Design
Production Coordinator: Nathan Johanson
Indexer: Matthew Spence
Cover Designer: Archer Design
Cover Photographer: Dewitt Jones
SYBEX is a registered trademark of SYBEX Inc.
Network Press and the Network Press logo are trademarks of SYBEX Inc.
TRADEMARKS: SYBEX has attempted throughout this book to distinguish proprietary trademarks from
descriptive terms by following the capitalization style used by the manufacturer.
Every effort has been made to supply complete and accurate information. However, SYBEX assumes no
responsibility for its use, nor for any infringement of the intellectual property rights of third parties which
would result from such use.
The first edition of this book was published under the title Novell's Complete Encyclopedia of Networking
1995 SYBEX Inc.
Copyright 1996 SYBEX Inc., 2021 Challenger Drive, Alameda, CA 94501. World rights reserved. No part
of this publication may be stored in a retrieval system, transmitted, or reproduced in any way, including but
not limited to photocopy, photograph, magnetic or other record, without the prior agreement and written
permission of the publisher.
Library of Congress Card Number: 95-72476
ISBN: 0-7821-1829-1
Manufactured in the United States of America
10 9 8 7 6 5 4 3 2 1


Prev Page 7 Next

MWarranty

SYBEX warrants the enclosed CD-ROM to be free of physical defects for a period of ninety (90) days after
purchase. If you discover a defect in the CD during this warranty period, you can obtain a replacement CD at
no charge by sending the defective CD, postage prepaid, with proof of purchase to:
SYBEX Inc.
Customer Service Department
2021 Challenger Drive
Alameda, CA 94501
(800) 227-2346
Fax: (510) 523-2373
After the 90-day period, you can obtain a replacement CD by sending us the defective CD, proof of purchase,
and a check or money order for $10, payable to SYBEX.
MDisclaimer
SYBEX makes no warranty or representation, either express or implied, with respect to this medium or its
contents, quality, performance, merchantability, or fitness for a particular purpose. In no event will SYBEX,
its distributors, or dealers be liable for direct, indirect, special, incidental, or consequential damages arising
out of the use of or inability to use the software even if advised of the possibility of such damage.
The exclusion of implied warranties is not permitted by some states. Therefore, the above exclusion may not
apply to you. This warranty provides you with specific legal rights; there may be other rights that you may
have that vary from state to state.
MCopy Protection
None of the material on the CD is copy-protected. However, in all cases, reselling or making copies of these
programs without authorization is expressly forbidden.


Prev Page 8 Next


[In] a certain Chinese encyclopedia...it is written that animals are divided into:
(a) those belonging to the Emperor
(b) those that are embalmed
(c) tame ones
(d) suckling pigs
(e) sirens
(f) fabulous ones
(g) stray dogs
(h) those included in the present classification
(i) those that tremble as if mad
(j) innumerable ones
(k) those drawn with a very fine camelhair brush
(l) others
(m) those that have just broken the water pitcher
(n) those that look like flies from a long way off
Jorge Luis Borges


Prev Page 9 Next


Acknowledgments
As with the first edition, this book would never have been completed without the help of
many people. These people deserve thanks for all their efforts and energy. Guy Hart-Davis
convinced me that it was time for a revision and set me to work. Several people did splen-
did work during the production process: Kris Vanberg-Wolff, a veteran of the first edition,
worked on the revision until her planned departure for the calmer (and tastier) world of
cooking school. Maureen Adams, Laura Arendal, and Nathan Johanson took over the pro-
duction chores after Kris left. They did an excellent job, especially considering the short
notice and even shorter revision schedule. My heartfelt thanks to all these folks.
Mary Madden's technical reviews were always full of gentle, constructive corrections
and useful suggestions for improvements. Although I may not have been smart enough to
act on all of them, the suggestions have improved the book immensely-for which I'm very
grateful.
Kris Vanberg-Wolff's eagle eyes and infallible grammatical sense found and fixed my
awkward phrasings, stylistic inconsistencies, and grammatical aberrations. I shudder to
think what the book would have looked like without the benefit of these efforts.
As always, I'm very grateful to all the people who worked between and behind the
scenes to make this book, and also to those who created the compact disc. Thanks also to
the many people who sent me information about their products and who took the time
to answer my questions.
Finally, I dedicate this book to my wife Luanne and my daughter Molly-for all the joy
and fun they provide, during both work and play hours.


Prev Page 10 Next


Table of Contents
Introduction
ix
Entries (Listed Alphabetically)
1
Appendix A: Acronyms and Abbreviations
1113
Appendix B: Bibliography and Other Resources
1235
Index
1251


Prev Page 11 Next


Introduction
Introduction

Introduction
MWhat You'll Find in This Book
As in the first edition, I've tried to make this Encyclopedia a comprehensive source of informa-
tion about matters relating to networking. I've also tried to present the information in a clear
and useful manner.
This book contains comprehensive, straightforward summaries of the major concepts,
issues, and approaches related to networking. Networking is defined broadly to encompass
configurations ranging from a couple of connected computers just a few feet apart to a network
of several thousand machines (of all types and sizes) scattered around the world. You'll find
discussions of networking as it's done by servers and clients, managers and agents, peers, and
even over the telephone.
You probably won't find anything here that you can't find in other places. However, I don't
know of any other book or source that collects so much network-related information in one
place. To find all the information summarized here, you would need to check hundreds of
books, disks, articles, Web pages, or other documents.
Despite its hefty size, this encyclopedia just scratches the surface of what there is to know
about networking. After all, how complete can any book be if just the World Wide Web on the
Internet has over 10 million hypertext documents. I do think, however, that this book scratches
deeper than most other references you'll find.
This revised edition updates entries for concepts and technologies that change rapidly or
where there have been major developments. I've also added considerable material about the
Internet (and especially about the World Wide Web), since interest in this networking phenom-
enon is growing at an astounding pace.
As in the first edition, I've tried to cover concepts rather than making this a how-to book.
Thus, you won't learn how to install networks or run specific programs. However, you will
learn about different types of programs and what they do. For example, you can read about
browsers and how they make exploring the World Wide Web possible; you'll also learn about
programs such as network operating systems and how they differ from ordinary operating
systems.
Concepts, Not Instructions


Prev Page 12 Next


x
Introduction
This book was obsolete from the moment it was written. That's because nothing changes faster
than vocabulary in a field where there is money to be made. Since major breakthroughs and
advances are still happening in the area of networking, there are new network-related words
and concepts to be found in almost every issue of every computer magazine. If you include
acronyms and abbreviations, the speed with which the vernacular expands is even faster. For
example, the first edition of this book was published under a year ago and it was no trouble
finding almost 2,000 new entries for Appendix A.
Given the futility of even trying to stay completely up-to-date, I've chosen to focus on the
more enduring concepts and facts-those that provide the foundations and background that
underlie the constantly changing terminology. This makes the Encyclopedia more generally
useful and enduring.
While core networking concepts change very little, the core does grow. For example, ten years
ago there was much less need to know about wireless communications because there were
fewer wireless products, as well as less public interest in the technology. Because of such
progress, the body of essential fundamentals grows with each year.
I expect to update and add to the material in the book, and hope to make the Encyclopedia
always effective, comprehensive, and useful. Fortunately, an electronic medium makes it easier
to grow in this way.
If you need to find out something about networking, look for it in this book. If you find an
entry for the topic, we hope you'll be more informed after you've read it. On the other hand,
if you can't find the information you need, didn't understand it, or don't think you learned
what you should have, please drop us a line and tell us.
Also, if there are concepts or terms you would like to see included, please let us know. If you
can provide references, that would be helpful. Even under the best of circumstances, there's lit-
tle chance that you'll get a reply to individual queries. However, we will read your comments
and suggestions and will try to use them to improve future versions of the book.
An Anchor in an Ocean of Words
Helping the Book Grow


Prev Page 13 Next


&&
Symbols &
Numbers


Prev Page 14 Next


2
& (Ampersand)

Symbols & Numbers
M
& (Ampersand)
The ampersand is used to indicate special
characters in HTML (Hypertext Markup
Language) documents-that is, documents
for the World Wide Web. For example,
& specifies the ampersand character
(&); ö specifies a lowercase o with
an umlaut, or dieresis, mark ().
M
< > (Angle Brackets)
Angle brackets are used in pairs to surround
markup tags in HTML (Hypertext Markup
Language) documents for the World Wide
Web. For example,

indicates a para-
graph break; and indicate the start
and end of a section that is to be displayed
in boldface.
M
* (Asterisk)
In several operating systems, the asterisk
serves as a wildcard character: to represent
one or more characters, such as in a file
name or extension. For example, a* matches
act, actor, and and, but not band.
In pattern matching involving regular
expressions, the asterisk matches the occur-
rences of the single character immediately
preceding it. For example, ba*th matches
bth, bath, and baaaaath, but not bbath.
In e-mail and in other contexts that use
plain text, asterisks are sometimes used
around words or phrases to indicate em-
phasis. For example, "I *really* want
to emphasize the second word in this
sentence."
M
@ (At sign)
The at sign is used to separate the username
from domain specifiers in e-mail addresses.
For example, mels@golemxiv.mit.edu
would indicate someone with username mels
on a computer named golemxiv at MIT.
M
\ (Backslash)
In some operating systems, such as DOS,
OS/2, and NetWare, the backslash character
separates directory names or directory and
file names in a path statement. By itself, the
backslash represents the root directory in
these operating systems.
In various programming and editing con-
texts, the backslash is used to escape the
character that follows. For example, \n is an
escape code to indicate a newline character
in many operating environments.
M// (Double Slash)
In URLs (Uniform Resource Locators), dou-
ble slash characters separate the protocol
from the site and document names. For
example, if it existed,
http://examplehost.ucsc.edu/
filename.html
would refer to a file named filename.html
residing on the examplehost machine at the
University of California at Santa Cruz. To
get to this file, you would use a server that
supports the HTTP (Hypertext Transport
Protocol).


Prev Page 15 Next


4B/5B Encoding
3
M
(Mu)
Used as an abbreviation for the prefix micro,
as in sec for microsecond and m for
micrometer. This order of magnitude corre-
sponds to 2-20, which is roughly 10-6, or
one-millionth.
SEE ALSO
Order of Magnitude
M
. and .. (Period and Double Period)
In hierarchically organized directory sys-
tems, such as those used by UNIX, DOS,
and OS/2, . and .. refer to the current and
the parent directories, respectively. In pat-
tern matching involving regular expressions,
the . matches any single character, except a
newline character.
M
? (Question Mark)
In many operating systems, a question mark
serves as a wildcard character that repre-
sents a single character, such as in a file or
directory name.
M/ (Slash)
The slash (also known as a forward slash or
a virgule) separates directory levels in some
operating systems (most notably UNIX), in
addresses for gopher, and in URLs (Uniform
Resource Locators). For example, the fol-
lowing URL specifies the name and location
of a hypertext version of the jargon file,
which contains definitions for terms and
events that have helped define the computer
culture:
http://www.phil.uni-sb.de/fun/jargon/
index.html
In this URL, the file is named index.html,
and it is located in the /fun/jargon directory
on a machine in Germany (de).
In other operating systems, such as DOS,
OS/2, and NetWare, a slash is sometimes
used to indicate or separate command line
switches or options for a command.
M1Base5
The IEEE 802.3 committee's designation
for an Ethernet network that operates at
1 megabit per second (Mbps) and that
uses unshielded twisted-pair (UTP) cable.
This configuration uses a physical bus,
with nodes attached to a common cable.
AT&T's StarLAN is an example of a 1Base5
network.
SEE ALSO
10BaseX; 10Broad36
M4B/5B Encoding
4B/5B encoding is a data-translation scheme
that serves as a preliminary to signal encod-
ing in FDDI (Fiber Distributed Data Inter-
face) networks. In 4B/5B, every group of
four bits is represented as a five-bit symbol.
This symbol is associated with a bit pattern
that is then encoded using a standard signal-
encoding method, usually NRZI (non-return
to zero inverted).
This preprocessing makes the subsequent
electrical encoding 80 percent efficient. For


Prev Page 16 Next


4
5B/6B Encoding
example, using 4B/5B encoding, you can
achieve a 100 megabit per second (Mbps)
transmission rate with a clock speed of only
125 megahertz (MHz).
In contrast, the Manchester signal-
encoding method, which is used in Ethernet
and other types of networks, is only 50 per-
cent efficient. For example, to achieve a 100
Mbps rate with Manchester encoding, you
need a 200 MHz clock speed.
M
5B/6B Encoding
A data-translation scheme that serves
as a preliminary to signal encoding in
100BaseVG networks. In 5B/6B, every
group of five bits is represented as a six-bit
symbol. This symbol is associated with a bit
pattern that is then encoded using a stan-
dard signal-encoding method, such as NRZ
(non-return to zero).
M8B/10B Encoding
A data-translation scheme related to 4B/5B 10BaseF
encoding that recodes eight-bit patterns into
10-bit symbols. 8B/10B encoding is used, for
example, in IBM's SNA (Systems Network
Architecture) networks.
M
9-Track Tape
A tape storage format that uses nine parallel
tracks on 1/2-inch, reel-to-reel magnetic
tape. Eight tracks are used for data, and one
track is used for parity information. These
tapes are often used as backup systems on
minicomputer and mainframe systems; digi-
tal audio tapes (DATs) are more common on
networks.
M
10BaseX
The designations 10Base2, 10Base5,
10BaseF, and 10BaseT refer to various
types of baseband Ethernet networks.
10Base2 uses thin coaxial cable. This ver-
sion can operate at up to 10 megabits per
second (Mbps) and can support cable seg-
ments of up to 185 meters (607 feet). It is
also known as thin Ethernet, ThinNet, or
CheaperNet, because thin coaxial cable is
considerably less expensive than the thick
coaxial cable used in 10Base5 networks.
10Base5 uses thick coaxial cable. This ver-
sion is the original Ethernet. It can operate
at up to 10 Mbps and support cable seg-
ments of up to 500 meters (1,640 feet). It is
also known as thick Ethernet or ThickNet.
10BaseF is a baseband 802.3-based Ethernet
network that uses fiber-optic cable. This
version can operate at up to 10 Mbps.
Standards for the following special-
purpose versions of 10BaseF are being
formulated by the IEEE 802.3:
10BaseFP (fiber passive): For desktops
10BaseFL (fiber link): For intermediate
hubs and workgroups
10BaseFB (fiber backbone): For central
facility lines between buildings
10Base2
10Base5


Prev Page 17 Next


66-Type Punch-Down Block
5
10BaseT is a baseband 802.3-based Ethernet
network that uses unshielded twisted-pair
(UTP) cable and a star topology. This ver-
sion can operate at up to 10 Mbps. It is also
known as twisted-pair Ethernet or UTP
Ethernet.
BROADER CATEGOR Y
Ethernet
SEE ALSO
1Base5; 10Broad36; 100BaseT
M10Broad36
10Broad36 is a broadband, 802.3-based,
Ethernet network that uses 75-ohm coaxial
(CATV) cable and a bus or tree topology.
This version can operate at up to 10 mega-
bits per second (Mbps) and support cable
segments of up to 1,800 meters (about
6,000 feet).
A 10Broad36 network uses differential
phase shift keying (DPSK) to convert the
data to analog form for transmission.
Because of the encoding details, a
10Broad36 network actually needs
18 megahertz (MHz) for each channel:
14 MHz to encode the 10 Mbps signal and
4 MHz more for collision detection and
reporting capabilities.
In a 10Broad36 network, throughput is
10 Mbps in each direction-that is, a total
bandwidth of 36 MHz is needed. This band-
width can be provided in a single cable or in
two separate cables. A split-cable approach
uses half the cable for each direction, which
means the cable must have a 36 MHz band-
width. A dual-cable approach uses separate
cables for each direction, so that each cable
needs only an 18 MHz bandwidth.
BROADER CATEGORIES
Ethernet; Network, Broadband
SEE ALSO
1Base5; 10BaseX
M
56K Line
A digital telephone circuit with a 64 Kbps
bandwidth, but with a bandwidth of only
56 Kbps data, with the other 8 Kbps being
used for signaling. Also known as an ADN
(Advanced Digital Network) or a DDS
(Dataphone Digital Service) line.
M
64K Line
A digital telephone circuit with a 64 Kbps
bandwidth. Also known as a DS0 (digital
signal, level 0) line. When the entire 64 Kbps
are allocated for the data, the circuit is
known as a clear channel. This is in contrast
to a circuit in which 8 Kbps are used for
signaling, leaving only 56 Kbps for data.
M66-Type Punch-Down Block
A device for terminating wires, with the
possibility of connecting input and output
wires. This type of punch-down block can
handle wires with up to 25 twisted pairs.
The 66-type have generally been superseded
by 110-type punch-down blocks.
SEE ALSO
Punch-Down Block
10BaseT


Prev Page 18 Next

6
100BaseFX
M
100BaseFX
A 100BaseT basal type variant that runs
over multimode fiber-optic cable. Nodes on
a 100BaseFX network can be up to 2 kilo-
meters apart. This variant is also written
100Base-FX.
SEE
100BaseT
COMPARE
100BaseT4; 100BaseTX
M100BaseT
The general name for any of three 100 Mbps
Ethernet variants that have just been made a
standard by an IEEE 802.3 subcommittee
(802.3u). 100BaseT Ethernet is one of the
candidates trying to become the standard
100 Mbps Ethernet. This version was devel-
oped and proposed originally by Grand
Junction, in collaboration with several other
corporations.
The term fast Ethernet is often used for
this version. This is unfortunate, since that
term is also used to refer to any Ethernet
implementation that supports speeds faster
than the official 10 Mbps standard. To add
to the confusing terminology, a software
product (no longer available) was also
named fastEthernet.
100BaseT Ethernet retains Ethernet's
CSMA/CD (Carrier Sense Multiple Access/
Collision Detect) media access method-in
contrast to the 100BaseVG variant (now
officially, IEEE 802.12)-which is the other
major 100 Mbps Ethernet available.
The main differences between fast (100
Mbps) Ethernet and standard (10 Mbps)
Ethernet are:
I A 100BaseT Ethernet allows a much
shorter gap between signals.
I A 100BaseT Ethernet requires either
higher-grade cable or more wire pairs.
It can run at 100 Mbps speeds on
Category 3 or 4 cable-provided four
pairs are available; Category 5 cable
requires only two pairs.
I Currently, a 100BaseT Ethernet can
support a network that is only about
a tenth of the length allowed for an
ordinary Ethernet network. For net-
works that use copper (as opposed to
fiber-optic) cabling: Two nodes of a
100BaseT4 network can be no further
apart than 205 meters-regardless of
whether the nodes are next to each
other.
The following variants of 100BaseT
Ethernet have been defined:
100BaseFX: Runs over multimode fiber-
optic cable. Nodes on a 100BaseFX
network can be up to two kilometers
apart.
100BaseTX: Uses two wire pairs,
but requires Category 5 unshielded
or shielded twisted pair (UTP or
STP) wire.
100BaseT4: Can use category 3, 4, or 5
UTP cable. The T4 in the name comes
from the fact that four wire pairs are
needed: two for sending and two for
receiving.


Prev Page 19 Next

100BaseVG
7
In some configurations, fast and ordinary
Ethernet nodes can share the same network.
Fast Ethernet devices identify themselves as
such by sending a series of FLPs (fast link
pulses) at startup.
PRIMAR Y SOURCES
IEEE 802.3u committee publications
BROADER CATEGORIES
Ethernet
COMPARE
100BaseVG
M
100BaseT4
A 100BaseT Ethernet variant that can use
category 3, 4, or 5 unshielded twisted pair
(UTP) cable. The T4 means that four wire
pairs are needed: two for sending and two
for receiving. Two nodes of a 100BaseT4
network can be no further apart than 205
meters, regardless of whether the nodes are
next to each other. This variant is sometimes
written 100Base-T4.
SEE
100BaseT
COMPARE
100BaseTX; 100BaseFX
M100BaseTX
A 100BaseT Ethernet variant that uses two
wire pairs, but requires Category 5 UTP or
STP wire. Two nodes of a 100BaseTX net-
work can be no further apart than 205
meters-regardless of whether the nodes are
next to each other. This variant is sometimes
written 100Base-TX.
SEE
100BaseT
COMPARE
100BaseT4; 100BaseFX
M
100BaseVG
100BaseVG is a version of Ethernet devel-
oped by Hewlett-Packard (HP) and AT&T
Microelectronics, and is currently under
consideration by an IEEE 802.12 committee.
It is an extension of 10BaseT Ethernet that
will support transmissions of up to 100
megabits per second (Mbps) over voice-
grade (Category 3) twisted-pair wire. The
VG in the name stands for voice-grade.
100BaseVG Ethernet differs from ordinary
(10 Mbps) Ethernet in the following ways:
I Uses demand priority (rather than
CSMA/CD) as the media access
method.
I Can use ordinary (Category 3)
unshielded twisted-pair (UTP) cable,
provided that the cable has at least
four wire pairs. Ordinary Ethernet
needs only two pairs: one to send and
one to receive.
I Uses quartet signaling to provide four
transmission channels (wire pairs)
instead of just one. All wire pairs are
used in the same direction at a given
time.
I Uses the more efficient 5B/6B NRZ
signal encoding, as opposed to the
Differences from 10 Mbps Ethernet


Prev Page 20 Next

8
100BaseX
Manchester encoding scheme used by
ordinary Ethernet.
I For category 3 cable, a VG network
can be at most 600 meters from end to
end-and only 200 meters if all hubs
in the network are connected in the
same wiring closet. These values
increase by 50%-that is, to 900 and
300 meters, respectively-when cate-
gory 5 cable is used. For VG using
fiber-optic cable, the most widely sepa-
rated network nodes can be up to
5000 meters, or 5 kilometers, apart.
100BaseVG is designed to provide an easy
upgrade path from 10 Mbps Ethernet. An
upgrade requires two new components:
I A 100BaseVG network interface card
(NIC) for each node being upgraded.
This NIC replaces the 10 Mbps version
in the node.
I A 100BaseVG hub to replace the 10
Mbps hub. This type of hub is plug-
compatible with a 10 Mbps hub, so
that the upgrade requires simply
unplugging a node from one hub and
plugging it into the 100BaseVG hub.
This can all take place in the wiring
closet.
If you are already using twisted-pair
Ethernet cabling, you may not need any
new wiring, provided that the cable has four
wire pairs.
Upgrading to 100BaseVG
100BaseVG/AnyLAN is an extension
of 100BaseVG, developed as a joint effort
between Hewlett-Packard and IBM. This
version also supports the Token Ring archi-
tecture, and it can be used with either Ether-
net or Token Ring cards (but not both at the
same time or in the same network). Because
the demand priority access method can be
deterministic, the 100BaseVG/AnyLAN
architecture could handle isochronous
data-that is, data (such as voice or video)
that requires a constant transmission rate.
The 100VG-AnyLAN Forum is the advo-
cacy group for this Ethernet variant. This
consortium includes over 20 members,
including Apple, Compaq, and IBM.
100Base VG/AnyLAN is also known simply
as VG or AnyLAN.
BROADER CATEGOR Y
Ethernet
SEE ALSO
HSLAN (High-Speed Local-Area
Network)
COMPARE
100BaseT
M
100BaseX
100BaseX (sometimes written as 100
Base-X) is a function that translates bet-
ween the FDDI (Fiber Distributed Data
Interface)-based physical layer and the
CSMA/CD-based data-link layer in a 100
100BaseVG/AnyLAN


Prev Page 21 Next

3174
9
megabit per second (Mbps) Ethernet pro-
posed by Grand Junction Networks. The
term was used more generally to refer to a
100 Mbps Ethernet developed by Grand
Junction, among others. This proposed spec-
ification has since become known as Fast
Ethernet, and has been refined into three
variants:
I 100BaseFX, which runs over fiber-
optic cable
I 100BaseT4, which runs over
unshielded twisted pair (UTP) cable
rated at Category 3 or higher-pro-
vided there are four available wire
pairs
I 100BaseTX, which runs over
Category 5 UTP cable
These variants all use the standard
CSMA/CD (carrier sense multiple access/
collision detection) medium access scheme
used by classic Ethernet. (In contrast, the
100BaseVG variant proposed by Hewlett-
Packard and other companies uses a demand
priority access scheme.) Specifications and
standards for the Fast Ethernet versions
have been debated by the IEEE 802.3u sub-
committee, and were just approved in June
1995.
BROADER CATEGOR Y
Ethernet
SEE ALSO
Fast Ethernet
COMPARE
100BaseVG
M
100 Mbps Ethernet
Any of several proposed 100 Mbps imple-
mentations of the Ethernet network archi-
tecture. Three different approaches have
been proposed: 100BaseVG, 100BaseX, and
fastEthernet. These implementations differ
most fundamentally in the media-access
methods and types of cable they use.
M110-Type Punch-Down Block
A device for terminating wires, with the
possibility of connecting input and output
wires. This type of punch-down block has
generally replaced the older 66-type blocks
originally used by the telephone company.
SEE ALSO
Punch-Down Block
M
193rd Bit
In a T1 communications channel, a framing
bit that is attached to every group of 192
bits. These 192 bits represent a single byte
from each of the 24 channels multiplexed in
a T1 line.
SEE ALSO
T1
M
3174
A cluster control unit for the IBM 3270
family of display terminals.


Prev Page 22 Next

10
3270
M
3270
The 3270 designation is used for a line of
terminals, communications controllers, and
printers that are used with IBM mainframes.
The 3270 devices use synchronous commu-
nications protocols, either SDLC (Synchro-
nous Data Link Control) or BSC (Binary
Synchronous Communication), to communi-
cate with the host.
In order for a stand-alone PC to commu-
nicate with an IBM mainframe, it must have
an add-in board that enables the PC to emu-
late a 3270 terminal.
M3270 Data Stream
In IBM's SNA (Systems Network Architec-
ture) environment, a stream in which char-
acters are converted and/or formatted, as
specified through control characters and
attribute settings.
M3274
The designation for a cluster controller that
can serve as a front end for an IBM main-
frame host. Devices, such as 3270 terminals
or printers, communicate with the host
through this controller. The 3274 cluster
controllers have been replaced by 3174
establishment controllers in newer
configurations.
M3278
The designation for a popular IBM terminal
used to communicate with IBM mainframes.
M
3279
The designation for a color version of the
3278 terminal used to communicate with
IBM mainframes.
M3705
The designation for a computer that serves
as a data communications controller for
IBM's 370-series mainframes. The 3705 also
has ports for asynchronous access over
dial-up lines.


Prev Page 23 Next

AA


Prev Page 24 Next

12
AA (Auto Answer)
AM
AA (Auto Answer)
A modem feature in which the modem can
automatically respond to a call and establish
a connection.
MAAL (ATM Adaptation Layer)
The topmost of three layers defined for the
ATM network architecture. The AAL medi-
ates between the ATM layer and the various M
communication services involved in a trans-
mission.
SEE ALSO
ATM (Asynchronous Transfer Mode)
MAAR (Automatic Alternate Routing)
In X.25 and other networks, the process
by which network traffic is automatically
routed to maximize throughput, minimize
distance, or balance channel usage.
MABM (Asynchronous Balanced Mode)
In the ISO's HDLC (High-Level Data-Link
Control) protocol, an operating mode that
gives each node in a point-to-point connec-
tion equal status as senders and receivers.
M
ABP (Alternate Bipolar)
A signal-encoding method.
SEE ALSO
Encoding, Signal
M
Abstract Syntax
A machine-independent set of language ele-
ments and rules used to describe objects,
communications protocols, and other items.
For example, Abstract Syntax Notation One
(ASN.1) was developed as part of the OSI
Reference Model; Extended Data Represen-
tation (XDR) was developed as part of Sun
Microsystems' Network File System (NFS).
MAC (Access Control)
A field in a token ring token or data frame.
AC (Alternating Current)
AC (alternating current) is a power supply
whose polarity (direction of flow) switches
periodically. AC is the type of electrical
power supplied for homes and offices.
With AC, the actual amount of power
being supplied at any given moment depends
on where in the switching process you are.
When plotted over time, a "pure" AC power
supply produces a sine wave.
Not all countries use the same switching
rate. For example, in North America, the
current switches polarity 60 times per sec-
ond; in most European countries, the rate is
50 times per second. These values are indi-
cated as cycles per second, or hertz (Hz).
Thus, electrical power in the United States
alternates at 60 Hz.
Not all devices can use AC. In some cases
the AC power must be converted to direct
current (DC), which provides a constant
voltage level and polarity. All digital systems
(such as computers) must use DC.
COMPARE
DC (Direct Current)


Prev Page 25 Next

Access Control
13
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
M
AC (Application Context)
In the OSI Reference Model, AC (applica-
tion context) is a term for all the application
service elements (ASEs) required to use an
application in a particular context.
More specifically, in network manage-
ment, the AC provides the ground rules that
serve to define the relationship between two MAccess Control
applications during a temporary connection.
These ground rules will determine the types
of services that can be invoked during the
connection and also the manner in which
information will be exchanged. Such a con-
text is important for defining the systems
management services provided by a CMISE
(common management information service
element).
SEE ALSO
ASE (Application Service Element);
CMISE (Common Management
Information Service Element);
Acceptable Use Policy (AUP)
M
Acceptable Use Policy (AUP)
SEE
AUP (Acceptable Use Policy)
M
Acceptance Angle
In fiber optics, a value that measures the
range over which incoming light will be
reflected and propagated through the fiber.
The size of this angle depends on the relative
refractive indexes of the fiber core, the clad-
ding, and the surrounding medium (which is
generally air).
M
Acceptance Cone
In fiber optics, the three-dimensional analog
of an acceptance angle. The cone generated
by revolving the acceptance angle 360
degrees with the center of the fiber's core
as the cone's point.
An operating system uses access control to
determine the following:
I
How users or resources can interact
with the operating system
I What a specific user or group of users
may do when interacting with the
operating system
I Who can access a file or directory and
what that user can do after accessing it
I How system or network resources can
be used
At the lowest levels, hardware elements
and software processes can obtain limited
access to the system through mechanisms
such as interrupts or polling. For example,
low-level access to DOS is through IRQs
(interrupt request lines) and through soft-
ware interrupts, such as INT 21H, which
provide programs with access to DOS capa-
bilities and to certain hardware resources.
Access-control measures can be associ-
ated with users, files and directories, or
resources. When assigned to users or groups
of users, these control measures are known
as access rights, access privileges, trustee
rights, or permissions. When associated with
files and directories, the access-control


Prev Page 26 Next

14
Access Control Decision Function (ACDF)
elements are known as attributes or flags.
Resources and other system objects gener-
ally have an associated access control list
(ACL), which contains all the users who
may use the resource.
Access control is generally specified by a
system administrator or by the owner of a
particular file or resource. Some access privi-
leges are determined for users during net-
work configuration; others may be assigned
when the user logs on to a network or begins
a session with an operating system.
Access-control issues can be complex,
particularly if multiple operating environ-
ments are involved, as on an internetwork.
One reason is that operating environments
differ in the access-control measures they
support. Because there are overlaps, omis-
sions, and definition differences, mapping
access controls between environments may
be complicated.
SEE ALSO
Access Rights
M
Access Control Decision Function
(ACDF)
SEE
ACDF (Access Control Decision
Function)
M
Access Control Enforcement Function
(ACEF)
SEE
ACEF (Accces Control Enforcement
Function)
M
Access Control Information (ACI)
SEE
ACI (Access Control Information)
M
Access Network
A network attached to the trunk of a
backbone network. This type of connection
usually requires a gateway or a router,
depending on the types of networks that
comprise the backbone network.
MAccess Rights
Access rights are properties associated with
files or directories in a networking environ-
ment; also known as access privileges or
trustee rights. Access rights determine how
users and network services can access and
use files and directories. All networking
environments and operating systems use
some type of access rights settings to control
access to the network and its resources.
Access rights are similar to security
attributes, which specify additional proper-
ties relating to a file or directory. Security
attributes can override access rights. In gen-
eral, rights are assigned to a user for a spe-
cific file or directory. Attributes are assigned
to a file or directory and control access by
any user, regardless of that user's rights. The
set of rights a user has been assigned to a file
or directory is called his or her trustee
assignment.
The number of access rights is relatively
small. The terminology and particular com-
bination of rights vary from system to sys-
tem. For example, in Novell's NetWare 3.x
and 4.x, access rights may be associated


Prev Page 27 Next

Access Rights
15
a
b
c
d
e
f
g
h
i
j
klm
n
o
p
q
r
s
t
u
v
w
x
y
z
with directories or files or both, and a right
may apply to all the files in a directory or
only to individual ones. In NetWare 2.x,
rights apply only to directories. See the table
"Novell NetWare Access Rights" for
descriptions of the access rights associated
with NetWare.
The meaning or effect of a specific privi-
lege may also be system-dependent. For
example, in an AppleShare environment, the
following access privileges are defined:
I See Files, which allows a user to see,
open, and copy files.
I See Folder, which allows a user to see a
folder (but not necessarily the folder's
contents). If this privilege is not set, the
folder does not even appear on the
user's screen.
I
Make Changes, which allows a user to
change the contents of a file or folder.
Even drastic changes such as deletions
are allowed.
These AppleShare environment privileges
may be granted to any of the following:
I
Owner: The user who created (and,
hence, owns) the file or folder.
I
Group: The collection of users to
whom the privilege is granted. This
may be a single user.
I Everyone: All users with access to the
file server.
NOVELL NETWARE ACCESS RIGHTS
ACCESS RIGHT
USAGE ALLOWED
Access Control (A)
Create (C)
Erase (E)
File Scan (F)
Modify (M)
Read (R)
Supervisory (S)
Write (W)
Allows you to modify the trustee assignments and inherited rights mask (IRM) for a
file. With Access Control rights, you can grant other users any rights except Supervi-
sory rights.
Allows you to create subdirectories or files within a directory. Also allows you to
salvage a file if it is deleted.
Allows you to delete a file or directory.
Allows you to see a file or directory name when listing the parent directory.
Allows you to change the name and attributes of a file or directory.
Allows you to open and read a file.
Allows you to exercise all rights to a file or directory, including the right to grant
Supervisory privileges to the file or directory to other users. (This right does not
exist in NetWare 2.x.)
Allows you to open, edit, and save a file.


Prev Page 28 Next

16
Access Time
In UNIX, owners, groups, and others
may be granted read, write, or execute per-
missions for a file or a directory, as follows:
I Read access for a file allows a user
to read or display the contents of a
file. Read permission for a directory
means the user can generate a direc-
tory listing.
I Write access for a file means the user
can edit the file or redirect output to it. M
Write access for a directory allows the
user to create a file or a subdirectory.
I Execute access for a file allows the user
to use the file name as a command.
Execute permission for a directory
means the user can pass through the
directory to subdirectories.
When a single machine or network
includes more than one environment, there
must be a well-defined rule for assigning
and determining access rights. For example,
in NetWare for Macintosh, the NetWare
access rights supersede the AppleShare
access privileges.
Similarly, there are mechanisms for ensur-
ing that access rights are applied only as
broadly as intended. For example, NetWare
uses an Inherited Rights Mask (version 3.x)
or Inherited Rights Filter (version 4.x) to
specify which access rights for a directory
are also applicable in a subdirectory.
BROADER CATEGOR Y
Access Control
SEE ALSO
Attribute; IRM/IRF (Inherited Rights
Mask/Inherited Rights Filter)
M
Access Time
In hard-disk performance, the average
amount of time it takes to move the read/
write heads to a specified location and
retrieve data at that location. The lower the
value, the better the performance. Currently,
hard disks with access times of less than 15
milliseconds are common.
Accounting
A process by which network usage can be
determined and charges assessed for use of
network resources, such as storage, access,
and services. Accounting measures include
blocks read, blocks written, connect time,
disk storage, and service requests.
Most network operating systems include
an accounting utility or support an add-on
accounting package. For example, NetWare
3.11 has an accounting option in its
SYSCON utility.
MAccounting Management
One of five OSI network management
domains defined by the ISO and CCITT.
This domain is concerned with the adminis-
tration of network usage, costs, charges, and
access to various resources.
SEE ALSO
Network Management
M
Account Policy
In networking and other multiuser environ-
ments, a set of rules that determines whether
a particular user is allowed to access the sys-
tem and what resources the user may use. In


Prev Page 29 Next

ACF (Advanced Communications Function)
17
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
Windows NT Advanced Server, the account MACF (Advanced Communications
policy determines the way in which pass-
words may be used in a domain (a group of
servers with a common security policy and
database).
M
Accumaster Integrator
A network management program from
AT&T.
M
ACD (Automatic Call Distributor)
A device that automatically switches an
incoming call to the next available line.
M
ACDF (Access Control Decision
Function)
In open systems, a function that uses various
types of information, such as ACI (access
control information), and guidelines to
decide whether to grant access to resources
in a particular situation.
MACE (Adverse Channel Enhancement)
A modem-adjustment method that allows
the modem to compensate for noisy lines.
For example, the modem might lower the
operating speed.
MACEF (Access Control Enforcement
Function)
In open systems, a function that enforces
the decision made by the ACDF (access
control decision function).
Function)
ACF (Advanced Communications Function)
is the base name for several IBM software
packages that operate under IBM's SNA
(Systems Network Architecture). In some
cases, the programs are revisions or exten-
sions of older programs.
The following programs are included:
I ACF/NCP (Advanced Communica-
tions Function/Network Control Pro-
gram): Resides in a communications
controller. It provides and controls
communications between the host
machine and the network devices.
I ACF/TCAM (Advanced Communica-
tions Function/Telecommunications
Access Method): Serves as an ACF/
VTAM application and provides mes-
sage handling and other capabilities.
I ACF/VTAM (Advanced Communica-
tions Function/Virtual Telecommuni-
cations Access Method): Provides and
controls communications between a
terminal and host programs. ACF/
VTAM supersedes and adds capabili-
ties to the older VTAM software.
I
ACF/VTAME (Advanced Communi-
cations Function/Virtual Telecommu-
nications Access Method-Entry): An
obsolete program that has been super-
seded by ACF/VTAM.
BROADER CATEGOR Y
SNA (Systems Network Architecture)


Prev Page 30 Next

18
ACI (Access Control Information)
M
ACI (Access Control Information)
In the CCITT's X.500 directory services
model, any information used in controlling
access to a file or directory.
M
ACID (Atomicity, Consistency,
Isolation, and Durability)
In transaction processing (TP), the attributes
that are desirable for a transaction.
M
ACK
In telecommunications, a control character
that indicates that a packet has been
received without an error. In certain net-
work architectures, ACK is the name for a
frame that sends such an acknowledgment.
The ASCII ACK character has value 6.
M
ACL (Access Control List)
In some networking environments, the ACL
is a list of services available on a network,
along with the users and devices that are
allowed to use each service. This list pro-
vides one way to control access to network
resources.
In NetWare Directory Services (NDS),
each object in the directory has a property
called the ACL, which lists all the other
objects that have trustee assignments (rights)
to that object.
MACONSOLE
In Novell's NetWare 3.x, ACONSOLE is a
utility that allows a network supervisor to
access a server through a modem from a
workstation, and to manage the server from
this workstation. In NetWare 3.x, the
RCONSOLE utility provides the same func-
tion across a direct connection.
In NetWare 4.x, RCONSOLE was
updated to add ACONSOLE's asynchro-
nous capability, and ACONSOLE was
removed.
BROADER CATEGOR Y
NetWare
M
ACS (Asynchronous Communications
Server)
An ACS is usually a dedicated PC or expan-
sion board that provides other network
nodes with access to any of several serial
ports or modems. The ports may be con-
nected to mainframes or minicomputers.
To access a modem or a port, the work-
station user can run an ordinary communi-
cations program in a transparent manner.
However, in order for this to work, one of
the following must be the case:
I The communications program
must include a redirector to route
the communication process to the
appropriate ACS.
I The workstation must have a special
hardware port emulation board
installed, which takes up one of the
workstation's expansion slots. In this
case, the communications package
does not need any special rerouting
capabilities.
I The user must run a redirection
program before starting the commu-
nications package. To work with a


Prev Page 31 Next

Adapter
19
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
software-based redirector, the commu-
nications package must be able to use
DOS interrupt INT 14H. Unfortu-
nately, many communications pro-
grams bypass this interrupt to access
the UART (universal asynchronous
receiver/transmitter) directly for faster
operation.
BROADER CATEGOR Y
Server
M
ACSE (Association Control Service
Element)
In the OSI Reference Model, an application-
level service that establishes the appropriate M
relationship between two applications, so
that they can cooperate and communicate
on a task, such as exchanging information.
M
Active
When used to describe hardware or a con-
figuration, active generally means that the
hardware does some signal processing-
cleaning, boosting, or both. For example,
an active hub boosts and cleans a signal
before passing it on.
M
Active Hub
In an ARCnet network, a component that
makes it possible to connect additional
nodes to the network and also to boost
signals that go through the hub.
SEE ALSO
Hub
MActive Link
In an ARCnet network, a box used to con-
nect two cable segments when both cable
segments have high-impedance network
interface cards (NICs) connected.
M
Active Star
A network configuration in which the
central node of a star topology cleans
and boosts a signal.
SEE ALSO
Topology, Star
ACU (Autocall Unit)
A device that can dial telephone numbers
automatically.
M
AD (Administrative Domain)
In the Internet community, a collection of
nodes, routers, and connectors that is man-
aged by a common administrator, such as an
organization or a company.
M
Adapter
A board that plugs into an expansion bus,
and that provides special capabilities, such
as video, fax, modem, network access, and
so on. Besides functionality, adapters are
distinguished by the width of the data bus
between the adapter and the PC. Adapters
may have 8-, 16-, or 32-bit connections.


Prev Page 32 Next

20
ADC (Analog-to-Digital Converter)
M
ADC (Analog-to-Digital Converter)
A device that converts an analog signal to
digital form.
M
ADDMD (Administrative Directory
Management Domain)
In the CCITT's X.500 directory services
model, a collection of directory system
agents (DSAs) under the control of a single
authority.
SEE ALSO
DSA (Directory System Agent)
MAddress
An address is a value used to specify a loca-
tion. The location may be an area of local or
shared memory, or it may be a node or other
device on a network.
Several types of addresses are distinguished
for network locations. The type of address
used in a particular context depends partly
on which protocol or device is creating the
address. Address information may be main-
tained in any of several ways, such as in
look-up tables or directories.
Some common types of network-related
addresses are hardware, network, node,
Internet, and e-mail (electronic mail). There
are other types of addresses, and not all
types of addresses are used in the same
conceptual model. Devices that connect
networks or network segments generally
get network and/or node addresses on
each network they connect.
A hardware address, also known as a physi-
cal address or a MAC address, is a unique
numerical value assigned to a network inter-
face card (NIC) during the manufacturing
process or by setting jumpers or switches
during network installation. One part of this
address is assigned to the manufacturer by
the IEEE (Institute of Electronics Engineers)
and is common to all components from that
manufacturer; the second part of the hard-
ware address is a unique value assigned by
the hardware manufacturer.
A network address is an arbitrary value that
is assigned identically to each station in a
particular network. As long as there is only
a single network, this value is automatically
unique. If two or more networks are con-
nected, each must have a different network
address. If a station (for example, a server)
connects to two networks, that station will
have two different network addresses.
A network address is also known as a net-
work number or an IPX external network
number.
In addition to a common network address,
each station in a network has a unique node
address. This value identifies a particular
node, or more specifically, the NIC assigned
to each node, in a particular network. This
address is also known as a node number or
station address.
When specified as a source or destination,
a network server or workstation may be
Network-Related Addresses
Hardware Address
Network Address
Node Address


Prev Page 33 Next

Address
21
a
b
c
d
e
f
g
h
i
j
klmnopqrstuvwxyz
identified by a network and a node address
or by a hardware address.
The node addresses for Ethernet cards are
factory-set, and no two cards have the same
number. The node addresses for ARCnet
and Token Ring cards are set by changing
jumpers or switches on the cards. If a node
contains two NICs, the node will have two
different network addresses.
An internal address is a unique value that
specifies a node with respect to a particular
server in a network, which is useful in net-
works that have multiple servers. This is a
logical address. Only certain network oper-
ating systems, such as NetWare, support
internal addresses.
See the figure "Examples of network
addresses" for an illustration of the kinds of
addresses discussed so far.
An Internet address is a network-layer
address that uniquely identifies a node on
an internetwork or on the Internet. This type
of address uses four bytes of storage, and
it is generally represented as four decimal
values separated by decimal points, as in
12.34.56.78. Certain bits from an Internet
Internal Address
Internet Address
EXAMPLES OF NETWORK ADDRESSES


Prev Page 34 Next

22
Address Bus
address can be masked to identify a subnet-
work that contains some of the nodes in the
internetwork.
Special protocols, such as the Address
Resolution Protocol (ARP), are used to con-
vert from an Internet to a hardware address;
other programs, such as the Reverse ARP
(RARP), convert from a hardware to an
Internet address.
An e-mail (electronic mail) address is
an application-layer address that identifies
a user's mailbox location in a message-
handling system. These addresses have
little in common with the other types of
addresses mentioned; however, the e-mail
address must be associated with the station's MAddress Bus
network and node address or with its hard-
ware address in order for messages to be
transferred from a sender to a receiver.
Several different formats are used for mem-
ory addresses in personal computers: flat
address space, segmented address, and
paged address.
An address in a flat address space is a simple
numerical value in the range between 0 and
the highest address value. For example, in a
machine with 1 megabyte of memory, the
addresses range from 0x00000 to 0xfffff.
An address in a segmented address space
consists of a segment and an offset value.
E-Mail Address
Memory-Related Addresses
Flat Address Space
Segmented Address
The segment value represents a (usually 16-
byte) location that is aligned on a paragraph
boundary. The offset value represents the
number of bytes to shift from this segment
address. DOS uses segmented addresses.
Certain types of address space actually con-
sist of two types of values. For example, in
expanded memory, locations in a special
set of chips, and hence, in a special set of
addresses, are mapped into special memory
buffers. These buffers are broken into pages
of a specific size.
Virtual memory also uses paged
addresses.
An address bus is the electrical signal lines
over which memory locations are specified.
Each line carries a single bit, so the number
of lines on the bus determines the number of
possible addresses:
I 20 lines allow access to 1 megabyte
(MB) of memory. Examples include
Intel's 8086 and 8088 processors.
I 24 lines provide access to 16 MB.
Examples include Intel's 80286 and
Motorola's 68000 processors.
I 32 lines provide access to 4 gigabytes
(AB). Examples include Intel's 80386,
80486, and Pentium; and Motorola's
68020 and later processors.
I 64 lines provide access to 16 exabytes
(EB). (An exabyte is a billion billion,
Paged Address


Prev Page 35 Next

Administration
23
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
or a quintillion, bytes.) Digital Equip-
ment Corporation's Alpha APX chip is
an example of a 64-bit address bus.
M
Address Mask
In the IP (Internet Protocol) addressing
scheme, a group of selected bits whose val-
ues identify a subnetwork; also known as a
subnet mask. All the members of that sub-
network share the same mask value. Using
an address mask makes it easier for the sys-
tem to reference a member of a particular
subnet.
M
Address Resolution
The process of mapping one type of address
to another; specifically, mapping a network
(local) address to a hardware-dependent
address. The most widely used method
of address resolution is the Address Resolu-
tion Protocol (ARP) or a variation of that
protocol.
M
Adjacent Channel
A frequency band immediately before or
after the current channel. For example, a
channel between 100 MHz and 500 MHz
and a channel between 700 MHz and 900
MHz are both adjacent to the channel
between 500 MHz and 700 MHz.
M
ADMD (Administration Management
Domain)
In the CCITT's X.400 Message Handling
System (MHS) model, an ADMD (Adminis-
tration Management Domain) is a network
or network section operated by the CCITT
(Consultative Committee for International
Telegraphy and Telephony) or a national
PTT (Post, Telegraph, and Telephone). Spe-
cific examples of ADMDs include MCImail
and AT&Tmail in the United States; British
Telecom Gold400mail in Britain.
ADMDs are public carriers, unlike
PRMDs (private management domains),
which are run by private organizations or
companies. In accordance with CCITT
guidelines, ADMDs handle any interna-
tional connections; PRMDs communicate
through a local ADMD. ADMDs can con-
nect PRMDs, but a PRMD cannot connect
ADMDs. Because all ADMDs run under the
auspices of CCITT, the conglomeration of
ADMDs in the world forms the backbone
for a global X.400 network.
BROADER CATEGORIES
MD (Management Domain); X.400
COMPARE
PRMD (Private Management Domain)
M
Administration
Administration involves the management
and maintenance of a computer system,
network, or environment.
An administrator's responsibilities may
be grouped into the following general
categories:
Configuration management: Handling
tasks such as user accounts, hardware
settings, access rights, and security.
Data-flow management: Monitoring
performance, managing memory
Administrative Tasks


Prev Page 36 Next

24
Administration Management Domain (ADMD)
and resources, making sure applica-
tions and data files are accessible, and
generally ensuring that data is flowing
properly.
Hardware maintenance: Installing, main-
taining, and diagnosing hardware
components.
Software maintenance: Installing applica-
tions and other software, software
version control, bug reporting and
resolution, and so on.
Help: Training users, providing documen-
tation for using the system resources
and applications, and offering other
support.
Various levels of administration are distin-
guished, including the following:
System: Refers to a particular division
in a company or a particular type of
hardware, such as mainframes or data-
base servers. System administration
responsibilities do not necessarily
involve networking issues; that is, a
system administrator may or may not
need to attend to issues relating to the
connections between machines, as well
as to the machines themselves.
Network: Usually refers to a LAN (local-
area network), but may encompass
machines in a larger range, provided
these machines are all connected by a
common architecture. In addition to
the individual machines, a network
administrator must keep track of the
connections between the machines.
Levels of Administration
Internetwork: Refers to multiple net-
works. Some or all of these networks
may use different architectures. An
internetwork administrator should be
able to assume that any subnetworks
are under the control of network
administrators, so that the internet-
work administrator can concentrate
on the connections between networks
rather than those between machines.
MAdministration Management Domain
(ADMD)
SEE
ADMD (Administration Management
Domain)
M
Administrative Domain (AD)
SEE
AD (Administrative Domain)
M
Advanced Function Printing (AFP)
SEE
AFP (Advanced Function Printing)
M
Advanced Intelligent Network (AIN)
SEE
AIN (Advanced Intelligent Network)
MAdvanced Mobile Phone Service
(AMPS)
SEE
AMPS (Advanced Mobile Phone Service)


Prev Page 37 Next

Agent
25
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
M
Advanced Research Projects Agency
(ARPA)
SEE
ARPA (Advanced Research Projects
Agency)
M
Advantage Networks
Advantage networks represent a networking
strategy from Digital Equipment Corpora-
tion (DEC), designed to add support for
protocols such as the TCP/IP suite to
DEC's OSI-compliant DECnet Phase V
architecture.
MAdverse Channel Enhancement (ACE)
SEE
ACE (Adverse Channel Enhancement)
M
Advertising
The process by which a network service
makes its presence and availability known
on the network. For example, Novell Net-
Ware services use the SAP (Service Advertis-
ing Protocol).
M
AE (Application Entity)
In the OSI Reference Model, an entity (pro-
cess or function) that runs all or part of an
application. An AE may consist of one or
more application service elements (ASEs).
MAFI (Authority and Format Identifier)
In the OSI Reference Model, part of the
address for the network-layer service access
point (NSAP). The AFI portion specifies the
authority, or administrator, that is allocating
the IDI (initial domain identifier) values. The
AFI also specifies the format of the IDI and
the DSP (domain specific part), which are
other parts of the NSAP address.
M
AFP (Advanced Function Printing)
In IBM's SAA (Systems Applications Archi-
tecture) environments, the ability to print
text and images; that is, to use all points
addressable (APA) printers.
M
AFT (Application File Transfer)
In the International Standardized Profile
(ISP) grouping, a prefix that identifies FTAM
(file transfer, access, and management) pro-
files. For example, AFT11 represents basic
file transfer.
M
Agent
In general, an agent is a program that can
perform a particular task automatically,
when appropriate or upon request by
another program. An agent is commonly
used to provide information to an applica-
tion, such as a network management pro-
gram. An agent may be machine- or
function-specific.
The following are some of the agents that
are found in networking-related contexts:
I In a client-server networking model,
an element that does work on behalf of
a client or a server application. For
example, in Novell's SMS (storage
management system) backup architec-
ture, a special backup agent, called a
TSA (target service agent), is loaded on


Prev Page 38 Next

26
Aging
every node that you want to back up
from a centralized location. The agent
allows the central backup program to
access and back up the data on that
node.
I In an IBM Token Ring architecture, an
element on the network interface card
that monitors certain aspects of the
node and ring performance, and that
reports this information to a network
management program or to a Ring
Error Monitor (REM).
I In network management and mon-
itoring, a terminate-and-stay-resident
(TSR) program that runs on a work-
station to monitor activity and report
this to a network management
program.
The data collected by an agent is orga-
nized and processed by an agent handler. In
network management, an agent handler may
organize and analyze data concerning some
network function or component.
MAging
A process by which old items or table entries
are removed in a systematic manner, such as
first in, first out. This process serves both to
update such tables and to speed up access.
M
AI (Authentication Information)
In network security, information used to
determine whether a user is legitimate and
authorized to access the system.
M
AIM (Analog Intensity Modulation)
In communications using light (rather than
electrical) signals, a modulation method in
which the intensity of the light source varies
as a function of the signal being transmitted.
M
AIN (Advanced Intelligent Network)
In telecommunications, the name for a
sophisticated digital network of the future.
MAIS (Alarm Indication Signal)
A signal used in the OSI network manage-
ment model and also in broadband ISDN
networks to indicate the presence of an
alarm or error somewhere on the network.
M
AL (Application Layer)
The topmost of the seven layers in the OSI
Reference Model.
SEE ALSO
OSI Layer
MAlarm
In various network environments, particu-
larly network management, an alarm is a
signal used to indicate that an abnormality,
a fault, or a security violation has been
detected. Alarms may be distinguished by
type, such as performance, fault, or security,
and also by the severity of the event that
caused the alarm.
At one extreme are critical events that
represent immediate threats to continued
network operation; for example, when a
crucial LAN (local-area network) node or a


Prev Page 39 Next

Algorithm
27
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
server goes down. In some network manage-
ment environments, such critical alarms may
trigger automatic response by the network
management package.
At the other extreme are events that are
not currently serious, but that may eventu-
ally become serious enough to threaten net-
work operation; for example, when network
traffic is getting close to the network's band-
width limit. Such events generally do not
require immediate correction but should be
monitored.
M
Alarm Indication Signal (AIS)
SEE
AIS (Alarm Indication Signal)
MAlarm Reporting Function (ARF)
SEE
ARF (Alarm Reporting Function)
MAlert
In network management, an alarm sent by
an agent to the administrator. An alert
reports that a problem has arisen or that a
threshold has been reached.
MAlgorithm
An algorithm is a predefined set of instruc-
tions for accomplishing a task. An algorithm
is guaranteed to produce a result in a finite
amount of time. Algorithms are used in
many ways in networking. For example,
there are hashing algorithms for finding file
names in a directory and timing algorithms
for deciding how long to wait before trying
to access a network.
In most cases, the algorithms are of little
interest to either the casual or intense net-
work user. However, several algorithms have
escaped from behind the scenes and have
actually become items in marketing litera-
ture and other product discussions. The
following are a few of the better-known
algorithms:
Auto-partition: An algorithm by which a
repeater can automatically disconnect
a segment from a network if that seg-
ment is not functioning properly. This
can happen, for example, when a bro-
ken or unterminated cable causes too
many collisions. When the collisions
have subsided, the network segment
can be reconnected.
Bellman-Ford: An algorithm for finding
routes through an internetwork. The
algorithm uses distance vectors, as
opposed to link states. The Bellman-
Ford algorithm is also known as the
old ARPAnet algorithm.
Distance-vector: A class of computation-
intensive routing algorithms in which
each router computes the distance
between itself and each possible desti-
nation. This is accomplished by com-
puting the distance between a router
and all of its immediate router neigh-
bors, and adding each neighboring
router's computations for the distances
between that neighbor and all of its
immediate neighbors. Several com-
monly used implementations are
available, such as the Bellman-Ford
algorithm and the ISO's Interdomain
Routing Protocol (IDRP).


Prev Page 40 Next

28
Alias
Hot potato: In networks, a routing algo-
rithm in which a node routes a packet
or message to the output line with the
shortest queue.
Link-states: A class of routing algorithms
in which each router knows the loca-
tion of and distance to each of its
immediately neighboring routers, and
can broadcast this information to all
other routers in a link state packet
(LSP). If a router updates its LSP, the
new version is broadcast and replaces
the older versions at each other router.
The scheme used to distribute the LSP
greatly influences the performance of
the routers. These types of algorithm
are an alternative to distance-vector
algorithms; rather than storing actual
paths, link-state algorithms store the
information needed to generate such
paths. The ISO's open shortest path
first (OSPF) algorithm is an example
of a link-state algorithm.
Spanning-tree: An algorithm that is used
to compute open paths (paths without
loops) among networks. The algorithm
can generate all such paths and select
one. If that path becomes inoperative
because a node has gone down, the
algorithm can find an alternate path.
This type of algorithm is used by
bridges to find the best path between
two nodes in different networks, and
to ensure that no path loops occur in
the internetwork. This algorithm is
defined in the IEEE 802.1 standard.
M
Alias
In a computer environment, a name that
represents another, usually longer, name. In
NetWare Directory Services (NDS), an alias
is an object in one part of the Directory
tree that points to the real object, which is
located in a different part of the tree. Users
can access the real object through the alias.
M
Alignment Error
In an Ethernet or other network, an error
in which a packet has extra bits; that is, the
packet does not end on byte-boundaries and
will have invalid CRC (cyclic redundancy
check) values. An alignment error may be
caused by a faulty component, such as a
damaged network interface card (NIC),
transceiver, or cable.
M
Allocation Unit
In Novell's NetWare, areas that are used
to store information from files and tables.
Two types of storage are distinguished:
blocks, which are used to store data on disk,
and buffers, which hold data in RAM
temporarily.
SEE ALSO
Block; Buffer, Fiber-Optic Cable; Buffer,
Memory.
M
Alternate Mark Inversion (AMI)
SEE
AMI (Alternate Mark Inversion)


Prev Page 41 Next

AMP (Active Monitor Present)
29
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
M
Alternate Route Selection (ARS)
SEE
ARS (Alternate Route Selection)
MAlternate Routing
This term describes the use of an alternative M
communications path, such as a telephone
connection, when the primary one is not
available.
M
AM (Accounting Management)
In network management, a function for
gathering performance and usage informa-
tion from a network.
M
AM (Active Monitor)
In a token ring network, the node that is
responsible for creating, passing, and main-
taining the token. The performance of the
AM is monitored constantly by standby
monitors (SMs) to ensure that the token-
passing process is not interrupted.
M
AME (Asynchronous Modem
Eliminator)
An AME, also known as a null modem, is a
serial cable and connector with a modified
pin configuration (compared to an ordinary
RS-232 cable). This cable enables two com-
puters to communicate directly; that is,
without modems as intermediaries.
M
American National Standards Institute
(ANSI)
SEE
ANSI (American National Standards
Institute)
America Online (AOL)
SEE
AOL (America Online)
M
AMF (Account Metering Function)
In the OSI network management model, the
function that keeps track of every user's
resource usage.
M
AMH (Application Message Handling)
In the International Standardized Profile
(ISP) model, the prefix used to identify MHS
(Message Handling System) actions.
M
AMI (Alternate Mark Inversion)
A signal-encoding scheme in which a 1 is
represented alternately as positive and nega-
tive voltage, and 0 is represented as zero
voltage. It does not use transition coding,
but can detect noise-induced errors at the
hardware level.
SEE ALSO
Encoding, Signal
M
AMP (Active Monitor Present)
In token ring networks, a packet issued
every 3 seconds by the active monitor (AM)


Prev Page 42 Next

30
Amplifier
on the ring to indicate that the AM is work-
ing and is still in charge.
MAmplifier
A device for boosting an analog signal. The
same service is provided by a repeater for
digital signals.
M
Amplitude
The magnitude, or level, of a signal. For
an electrical signal, it is expressed in volts
(voltage) or amperes (current). In computer
contexts, current is more likely to be
expressed in milliamperes.
M
AMPS (Advanced Mobile Phone
Service)
A cellular telephone service. AMPS is a wire-
less analog communications service that
operates in the 825 to 890 megahertz range.
M
Analog Communication
A telecommunications system that uses
analog (that is, continuous, sinusoidal)
signals to represent information. An exam-
ple of an analog communication system is
the classic voice-based telephone system
(which is being replaced by the newer, digital
systems).
MAnalog Intensity Modulation (AIM)
SEE
AIM (Analog Intensity Modulation)
M
Analog-to-Digital Conversion
The process of converting an analog signal
(one that can take on any value within a
specified range) to digital form. An analog-
to-digital converter (ADC) is a device that
converts an analog signal to digital form.
M
ANF (AppleTalk Networking Forum)
A consortium of developers and vendors
working to encapsulate AppleTalk in other
protocols; for example, within the TCP/IP
suite.
MANI (Automatic Number
Identification)
In ISDN and some other telecommunica-
tions environments, a feature that includes
the sender's identification number, such as
telephone number, in the transmission, so
that the recipient knows who is calling; also
known as caller ID.
M
Annex D
In frame-relay technology, a document that
specifies a method for indicating permanent
virtual circuit (PVC) status. The document is
part of the ANSI T1.617 standard.
MAnonymous FTP
On the Internet, a protocol that allows a
user to retrieve publicly available files from
other networks. By using the special user ID,
"anonymous" users can transfer files with-
out a password or other login credentials.
(FTP is an application-layer protocol in the
Internet's TCP/IP protocol suite.)


Prev Page 43 Next

Anti-Virus Program
31
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
M
Anonymous Remailer
An Internet service that can be used to hide
the origins of an e-mail message being sent
to someone. The anonymous remailer
removes any source address information
from a message, substitutes any specified
pen name, and then sends the message on
to the specified destination.
M
ANSI (American National Standards
Institute)
The United States representative in the ISO
(International Standardization Organiza-
tion). ANSI creates and publishes standards
for programming languages, communica-
tions, and networking. For example, the
standard for the FDDI network architecture
is ANSI X3T9.5.
MAnti-Virus Program
An anti-virus program is used for detecting
or removing a computer virus. An anti-virus
program looks for suspicious activity, such
as unnecessary disk access, attempts to inter-
cept a BIOS or other low-level call, and
attempts to format or delete files. In some
cases, the anti-virus program detects a pat-
tern characteristic of a particular virus.
Some anti-virus programs are TSR
(terminate-and-stay-resident) programs,
which monitor computer activity constantly,
looking for indications of a virus. In some
cases, these types of programs can be
extremely annoying and very processor
intensive. Users have been known to remove
an anti-virus TSR program from memory
out of frustration.
Other anti-virus programs are intended to
be run periodically. When they are run, the
programs look for the tell-tale signs (known
as signatures) of particular viruses. These
programs are minimally disruptive; on the
other hand, their effectiveness is directly
proportional to the frequency with which
they are used.
Because the coding for computer viruses
is constantly changing, anti-virus programs
must also be updated regularly. It is impor-
tant to test anti-virus programs thoroughly,
which means that every new release must be
tested. Make sure an anti-virus program per-
forms to your expectations before installing
it on a network. Some programs can eat up
a significant amount of working memory.
Recently, a very different (and, conse-
quently, very controversial) type of anti-
virus program has become available. InVirc-
ible, created by Zvi Netiv, is designed to
detect viruses that have already infected a
system, and to clean these up. Rather than
looking for virus signatures, InVircible uses
expert system rules to look for behavior
characteristic of viruses: replication, use of
memory, attempts to attach to the anti-virus
program, etc. InVircible will even put out
"virus bait" to get an existing virus to try to
infect the bait.
BROADER CATEGOR Y
Data Protection
RELATED AR TICLE
Virus


Prev Page 44 Next

32
AOL (America Online)
M
AOL (America Online)
America Online is a commercial online ser-
vice like CompuServe and Prodigy. AOL
supports both DOS and Windows users, and
provides a range of services (mail, news, ref-
erence, financial, entertainment, Internet
access, etc.). Users pay a flat monthly fee,
which allows a limited number of free hours.
Additional hours are billed at a predeter-
mined rate. AOL's graphical interface is
highly regarded-in fact, Apple has licensed
the interface technology for use in Apple's
eWorld interface. AOL provides a very com-
prehensive set of access opportunities to the
Internet.
FOR INFORMATION
Call AOL at 800-827-6364
M
AOM (Application OSI Management)
In the International Standardized Profile
(ISP) model, the prefix for functions and
services related to network management.
M
AOW (Asia and Oceania Workshop)
One of three regional workshops for imple-
menters of the OSI Reference Model. The
other two are EWOC (European Workshop
for Open Systems) and OIW (OSI Imple-
menters Workshop).
M
AP (Application Process)
In the OSI Reference Model, a program that
can make use of application layer services.
Application service elements (ASEs) provide
the requested services for the AP.
M
APD (Avalanche Photodiode)
A detector component in some fiber-optic
receivers. The APD converts light into elec-
trical energy. The "avalanche" refers to the
fact that the detector emits multiple elec-
trons for each incoming photon (light
particle).
MAPDU (Application Protocol Data
Unit)
A data packet at the application layer; also
called application-layer PDU.
SEE ALSO
OSI Reference Model
M
API (Application Program Interface)
An abstract interface to the services and pro-
tocols offered by an operating system, usu-
ally involving a published set of function
calls. Programmers and applications can
use the functions available in this interface
to gain access to the operating system's
services.
MAPIA (Application Program Interface
Association)
A group that writes APIs for the CCITT's
X.400 Message Handling System (MHS).
M
APPC (Advanced Program-to-
Program Communications)
In IBM's SAA (Systems Application Archi-
tecture), APPC is a collection of protocols
to enable executing applications to commu-
nicate directly with each other as peers
(without intervention by a mainframe host).


Prev Page 45 Next

AppleTalk
33
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
APPC is defined at a level comparable
to the session layer in the OSI Reference
Model. It can be supported in various net-
working environments, including IBM's
SNA (System Network Architecture),
Ethernet, Token Ring, and X.25.
APPC/PC (Advanced Program-to-
Program Communications/Personal Com-
puters) is a PC-based version of APPC.
M
AppleDouble
In the Macintosh world, a file format that
uses separate files for the data and resource
forks that make up a Macintosh file. This
enables the files-or at least the data por-
tion-to be used on different platforms.
COMPARE
AppleSingle
M
AppleShare
A network operating system from Apple.
AppleShare runs on a Macintosh network
server, providing file and printer services.
AppleShare uses the AppleTalk protocol
suite to carry out its tasks.
SEE ALSO
AppleTalk
MAppleSingle
In the Macintosh world, a file format that
stores both a file's contents (data fork) and
its resources (resource fork) within a single
file. Because data and resources are mixed in
a proprietary format, such a file cannot be
used on other platforms.
COMPARE
AppleDouble
M
AppleTalk
AppleTalk is Apple's proprietary protocol
suite for Macintosh network communica-
tions. It provides a multilayer, peer-to-peer
architecture that uses services built into the
operating system. This gives every Macin-
tosh networking capabilities. AppleTalk can
run under any of several network operating
systems, including Apple's AppleShare,
Novell's NetWare for Macintosh, and Sun
Microsystems' TOPS.
AppleTalk was developed in the mid-
1980s with the goal of providing a simple,
portable, easy-to-use, and open networking
environment. To access such a network,
a user just needs to "plug in, log in, and
join in."
A newer version, Phase 2, was released in
1989. This version provided some new capa-
bilities and extended others.
AppleTalk is a comprehensive, layered envi-
ronment. It covers networking services over
almost the entire range of layers specified in
the OSI Reference Model. The figure "The
AppleTalk protocol hierarchy" shows the
organization of the AppleTalk layers, as well
as the protocols in the AppleTalk Protocol
Suite.
AppleTalk Layers


Prev Page 46 Next

34
AppleTalk
THE APPLETALK PROTOCOL HIERARCHY
Please register!


Prev Page 47 Next

AppleTalk
35
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
There are AppleTalk implementations for
the following network architectures at the
physical and data-link layers:
I Apple's 230 kilobit per second (Kbps).
I LocalTalk architecture. LocalTalk pro-
vides a media-access method and a
cabling scheme for AppleTalk. The
architecture uses twisted-pair cables
and RS-422 connections, allows nodes
to be separated by as much as 305
meters (1,000 feet), and can transmit
at up to 230.4 Kbps. The term Local-
Talk is sometimes used to refer to an
AppleTalk network.
I EtherTalk, Apple's implementation
of the 10 megabit per second (Mbps)
Ethernet architecture. Two versions
of EtherTalk exist. The earlier one,
EtherTalk Phase 1, is modeled on the
Blue Book Ethernet 2.0 (as opposed
to the version specified in the IEEE
802.3 documentation). Its successor,
Phase 2, is modeled on the IEEE 802.3
standard. Because these two variants
of Ethernet define packets somewhat
differently, Phase 1 and Phase 2 nodes
cannot communicate directly with
each other. EtherTalk has replaced
LocalTalk as the default networking
capability in newer Macintosh models.
I TokenTalk, Apple's implementation of
the token-ring architecture. AppleTalk
supports both the 4-Mbps version
specified by IEEE 802.5 and the 16-
Mbps version from IBM. The token-
ring architecture is supported only in
AppleTalk Phase 2.
Physical and Data-Link Layers
I FDDITalk, Apple's implementation
of the 100 Mbps FDDI architecture.
For each of these architectures, a Link
Access Protocol (LAP) is defined: LLAP for
LocalTalk, ELAP for EtherTalk, TLAP for
TokenTalk, and FLAP for FDDITalk.
All AppleTalk networks use the DDP (Data-
gram Delivery Protocol) at the network
layer, regardless of the architecture operat-
ing at the data-link layer. This protocol
makes a best effort at packet delivery, but
delivery is not guaranteed.
Note also the AARP (AppleTalk Address
Resolution Protocol) at this layer. The
AARP maps AppleTalk (network) addresses
to Ethernet or Token Ring (physical)
addresses.
For reliable packet delivery, the ADSP
(AppleTalk Data Stream Protocol) and
ATP (AppleTalk Transaction Protocol)
are available. Each of these protocols is
appropriate under different conditions.
The NBP (Name Binding Protocol) and
ZIP (Zone Information Protocol) help make
addressing easier. NBP associates easy-to-
remember names (used by users) with the
appropriate address.
ZIP is used mainly on larger networks or
internetworks, which are more likely to be
divided into zones. A zone is a logical group-
ing of nodes that together make up a subnet-
work. The concept of a zone was introduced
to allow for larger networks with more than
255 nodes, and also to make addressing and
routing tasks easier.
Network Layer
Higher Layers


Prev Page 48 Next

36
AppleTalk
Applications access an AppleTalk net-
work through the AFP (AppleTalk Filing
Protocol); they access printer services by
shipping PostScript files through the PAP
(Printer Access Protocol).
A few protocols make use of services
from more than one lower-level protocol.
For example, ZIP relies on ATP and DDP
services.
The following protocols make up the Apple-
Talk Protocol Suite (see the figure "The
AppleTalk protocol hierarchy," earlier in
this article):
AARP (AppleTalk Address Resolution
Protocol): A network-layer protocol
that maps AppleTalk (network)
addresses to physical addresses.
ADSP (AppleTalk Data Stream Protocol):
A session-layer protocol that allows
two nodes to establish a reliable con-
nection through which data can be
transmitted.
AEP (AppleTalk Echo Protocol): A
transport-layer protocol used to deter-
mine whether two nodes are connected
and both available.
AFP (AppleTalk Filing Protocol): A pre-
sentation/application-layer protocol
used by applications to communicate
with the network.
ASDSP (AppleTalk Safe Data Stream
Protocol): A session-layer protocol
that is similar to ADSP but that pro-
vides additional security against
unauthorized use.
ASP (AppleTalk Session Protocol): A
session-layer protocol used to begin
and end sessions, send commands
from client to server, and send replies
from server to client.
ATP (AppleTalk Transaction Protocol): A
transport-layer protocol that can pro-
vide reliable packet transport. Packets
are transported within the framework
of a transaction (an interaction
between a requesting and a responding
entity {program or node}).
AURP (AppleTalk Update Routing
Protocol): A transport-layer routing
protocol that is similar to RTMP
(Routing Table Maintenance Proto-
col) but that updates the routing table
only when a change has been made to
the network.
DDP (Datagram Delivery Protocol): A
network-layer protocol that prepares
and routes packets for transmission on
the network.
LAP (Link Access Protocol): Works at
the data-link layer, converting packets
from higher layers into the appropriate
form for the physical transmission.
Each network architecture needs its
own LAP.
ELAP (EtherTalk Link Access Protocol):
The link-access protocol used for
Ethernet networks.
FLAP (FDDITalk Link Access Protocol):
The link-access protocol used for
FDDI networks.
AppleTalk Protocol Suite


Prev Page 49 Next

AppleTalk
37
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
LLAP (LocalTalk Link Access Protocol):
The link-access protocol used for
LocalTalk networks.
TLAP (TokenTalk Link Access Proto-
col): The link-access protocol used
for Token Ring networks.
ARAP (AppleTalk Remote Access
Protocol): A link-access protocol for
accessing the network from a remote
location over a serial line.
NBP (Name Binding Protocol): A
transport-layer protocol that associ-
ates device names with network
addresses. If the NBP is successful, this
binding process will be completely
transparent to the user.
PAP (Printer Access Protocol): A session-
layer protocol for creating a path from
the user or application to a printer.
RTMP (Routing Table Maintenance
Protocol): A transport-layer routing
protocol for moving packets between
networks.
ZIP (Zone Information Protocol): A
session-layer protocol used to help
find a node; for example, in a large
internetwork.
If installed, an AppleShare server runs
on top of these protocols at the uppermost
(application) layer. The AppleShare server
uses the AFP to provide centralized file shar-
ing for its clients, and can use the PAP to
provide printer sharing.
In AppleTalk networks, every node has an
official numerical address. In addition, a
node may be part of a named group of
nodes, which somehow belong together.
Each AppleTalk network is assigned a
unique network number, and each node in
that network is assigned this number. Pack-
ets addressed to a node on the network must
include the network number.
In addition to a network number, each
node has a node number that is unique
within that network. This is an 8-bit number
and can be any value between 1 and 254,
inclusive (0 and 255 are reserved as node
numbers). However, servers must have node
numbers within the range of 128 to 254,
and workstations must have numbers in
the 1 to 127 range.
A zone is a logical grouping of nodes. The
basis for the grouping can be any criterion
that is useful for a particular configuration,
as in the following examples:
I Geographical, such as all machines on
the second floor
I Departmental, such as all machines in
the marketing department
I Functional, such as all machines that
can provide access to printers
By restricting routing or searches to
machines in a particular zone, network traf-
fic and work can be reduced considerably.
Numbers and Zones
Network and Node Numbers
Zones


Prev Page 50 Next

38
AppleTalk
Accessing resources by zones also makes
it easier to determine what is available for
specific needs.
A node may belong to more than one
zone at the same time, or not be part of any
zone. A zone can cross network boundaries;
that is, a zone can consist of parts of two or
more different networks or include multiple
networks.
Phase 2, an updated version of AppleTalk,
was released in 1989. This version provides
several improvements over Phase 1, includ-
ing the following:
I Allows more than 254 nodes per
network
I Allows a network to be assigned more
than one network number
I Introduced the AppleTalk Internet
Router, which allows up to eight
AppleTalk networks to be connected
In AppleTalk Phase 2, a network can be
assigned a range of network numbers. A
particular node on this network can be asso-
ciated with any one number in this range. By
providing multiple network numbers for a
single network, it is possible to have more
than the 254 nodes allowed in a Phase 1 net-
work, because each network number can
support 253 (yes, 253) individual nodes.
Phase 2 AppleTalk
Network Numbering in Phase 2
When you are assigning number ranges,
a rough guideline is to assign one network
number for every 25 to 50 nodes. If you
expect a lot of growth, use a smaller num-
ber. For example, assigning two network
numbers for a 100-node network leaves
room for 406 additional nodes.
When a network is part of an internet-
work, there are several restrictions on what
can be connected and how. These restric-
tions concern routers and bridges, and the
networks they can connect, as follows:
I All routers connected to a particular
network must use the same network
number range for the interface with
that network. For example, if a router
thinks the network uses numbers
1,000 to 1,009, another router con-
nected to the same network cannot use
1,002 to 1,008.
I Routers must connect networks with
different number ranges that do not
overlap. This means that routers can-
not connect a network to itself and
that networks with overlapping net-
work numbers cannot interact with
each other.
I A bridge must connect network seg-
ments with the same number range.
The figure "Rules for connecting AppleTalk
Phase 2 internetworks" illustrates these
rules.


Prev Page 51 Next

Application
39
a
b
c
d
efghijklmn
o
p
q
r
s
t
u
v
w
x
y
z
M
AppleTalk Networking Forum (ANF)
SEE
ANF (AppleTalk Networking Forum)
MApplication
An application is a program that calls oper-
ating system services and performs work,
such as data creation or manipulation, for
the user. Applications may be stand-alone,
network-based, or part of an integrated
package.
A stand-alone application can execute only
one version of itself at a time and can sup-
port only a single user at a time. This type of
application executes on a single machine,
which may or may not be connected to a
network. Single-user versions of spread-
sheet, graphics, and database programs
are examples of stand-alone applications.
A network-based application executes on a
network and is aware of the network, which
Stand-Alone Applications
Network-Based Applications
RULES FOR CONNECTING APPLETALK PHASE 2 INTERNETWORKS


Prev Page 52 Next

40
Application Entity (AE)
means that it can use networking conven-
tions, elements, resources (such as print
spoolers and cache buffers), and devices
(such as printers, modems, and backup
devices).
This type of application can be used by
multiple users at the same time. Applications
differ in the number of allowable users and
in the measures taken to enforce restrictions
and to make sure users do not ruin other
users' data. Network and data protection
measures include the use of flags, access
rights, and lock-outs. These serve to help
ensure that data is used correctly, only as
needed, and with fair access to all users.
Network-based applications may execute
on a single machine or be distributed over
multiple machines. Client/server computing
is an example of a distributed arrangement
in which part of an application (the front
end) executes on the workstation to provide
an interface for the user, and another part
(the back end) executes on a server to do the
actual work, such as searching a database.
A network-based application may be
multiuser or multilaunch. Only one copy of
a multiuser application executes, but multi-
ple users can access files in this executing
program. A multilaunch application allows
multiple users to execute the program sepa-
rately but at the same time. In effect, each
user gets a private version of a multilaunch
application.
An integrated application is part of a collec-
tion, or suite, of programs. Ideally, these
programs complement each other in their
functionality and allow easy exchange of
Integrated Applications
data. Microsoft Office, Lotus SmartSuite,
and Borland Office are examples of such
integrated applications.
Users may access networks through or for
applications. For example, an application
may use a network resource or may need
to communicate with an application on
another machine. Or a user may log in to a
network with the specific intention of using
an application available on that network.
Regardless of the details, such network
accesses are through the topmost layer in the
OSI Reference Model: the application layer.
This layer provides users and programs with
an interface to the network. At this layer,
both the user and the application are iso-
lated from the details of network access and
communication.
M
Application Entity (AE)
SEE
AE (Application Entity)
Accessing Networks from Applications
SHARING DATA AMONG
APPLICATIONS
Separate applications can also communicate and
exchange data. Using pipes, in which the output
from one program is simply "piped" in as input to
another program is one of the simplest ways to
share data.
OLE (object linking and embedding) is a more
sophisticated method, which provides much
greater flexibility. OLE makes it possible for
updates to be carried over automatically to
whatever applications use the updated items.


Prev Page 53 Next

Archie
41
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
M
Application File Transfer (AFT)
SEE
AFT (Application File Transfer)
M
Application Layer
The topmost layer in the seven-layer OSI
Reference Model.
SEE ALSO
OSI Reference Model
M
Application Process (AP)
SEE
AP (Application Process)
M
Application Program Interface (API)
SEE
API (Application Program Interface)
M
Application Program Interface
Association (APIA)
SEE
APIA (Application Program Interface
Association)
M
Application Protocol Data Unit
(APDU)
SEE
APDU (Application Protocol Data Unit)
M
APPN (Advanced Peer-to-Peer
Networking)
APPN is a network architecture defined
within IBM's SAA (Systems Application
Architecture) environment. APPN allows
peer-to-peer communications between com-
puters without requiring a mainframe in the
network.
APPN is also supported within IBM's
SNA (Systems Network Architecture) envi-
ronment. Unlike standard SNA, however,
APPN supports dynamic routing of packets.
BROADER CATEGOR Y
SAA (Systems Application Architecture)
M
ARA (Attribute Registration
Authority)
In the X.400 Message Handling System
(MHS), the organization that allocates
unique attribute values.
MArchie
An Internet service that can find the location
of specified files based on the file's name or
description. An archie server gets its infor-
mation by using the FTP program to do a
listing of files on accessible servers and also
by getting file description information. Cur-
rently, archie servers have data about over
2.5 million files on over 1,000 servers.
Archie servers are scattered throughout
the Internet, and are accessible using services
such as telnet or gopher, through e-mail, or
by using archie client programs. Archie serv-
ers should be equivalent (except for minor
differences arising because not all servers are
updated at the same time), so selecting a
server is just a matter of convenience. See
the table "Example Archie Servers" for a
list of some of the available servers.


Prev Page 54 Next

42
Architecture
Once a connection has been established with
the archie server, various commands are
available. The following list summarizes
some useful ones:
help
Displays a list of available
commands.
manpage Displays the reference manual
for archie.
list
Displays a list of the anony-
mous STP servers whose con-
tents are listed in archie's
database. If this command is
EXAMPLE ARCHIE SER VERS
SERVERS
LOCATION
archie.ac.il
archie.au
archie.doc.ic.ad.uk
archie.edvz.uni-linz.ac.at
archie.funet.fi
archie.kr
archie.mcgill.ca
archie.ncu.edu.tw
archie.rediris.es
archie.rutgers.edu
archie.sura.net
archie.switch.ch
archie.th-darmstadt.de
archie.unipi.it
archie.univ-rennes1.fr
archie.unl.edu
archie.wide.ad.jp
Israel
Australia
United Kingdom
Austria
Finland
Korea
Canada (McGill
University)
Taiwan
Spain
USA (Rutgers
University)
(SURAnet is a service
provider)
Switzerland
Germany
Italy
France
USA (University of
Nebraska, Lincoln)
Japan
Useful Archie Commands
followed by a regular expres-
sion, the command displays
only the servers that match
the expression.
servers
Displays a list of all the avail-
able archie servers.
version
Displays the version number
of the archie server you're
querying. Such information
will come in handy if you
need to get help with the
program.
Various other commands and configu-
ration possibilities are available to make
archie more useful and more convenient
to use.
M
Architecture
Architecture is an amorphous term in the
area of networking. The term can refer to
both the physical layout (topology) of the
network and also the protocols (communi-
cation rules and data elements) used to
communicate.
Architecture can also refer to the basic
structure of a networking service, such as
a print service architecture. Used this way,
it generally indicates the overall scheme
of APIs (Application Program Interfaces),
agents, and so on, used to fit different pieces
of the service together.
You will hear references to network
architectures, such as ARCnet, Ethernet, and
Token Ring, which are all defined primarily
at the two lowest layers of the OSI model:
the physical and data-link layers. Each
architecture includes an implicit topology.


Prev Page 55 Next

ARCnet (Attached Resource Computer Network)
43
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
In the context of hardware, the term
refers to the manner in which a computer is
constructed. The architecture includes the
type of processor (for example, Intel 80x86
or Pentium, Motorola 680xx, or RISC chip)
and the type of bus that is used to transmit
data and other signals to the computer's
components and peripherals.
In the IBM PC world, which is currently
dominated by Intel processors, the three
major buses are ISA (Industry Standard
Architecture), EISA (Extended Industry
Standard Architecture), and MCA (Micro-
channel Architecture). However, two newer
bus designs-VL (VESA Local) and PCI
(Peripheral Component Interconnect)-are
growing in popularity and are likely to
become the dominant bus architectures.
SEE ALSO
Network Architecture
M
Archive
As a noun, a repository for data, applica-
tions, and so forth. These materials may be
master copies or regular backups of the cur-
rent hard disk contents. As a verb, the act of
backing up data files to provide a safe copy
in case of a disaster.
M
Archive Site
On the Internet, a node that provides access
to a collection of files.
M
ARCnet (Attached Resource
Computer Network)
ARCnet is a baseband network architec-
ture originally developed as a proprietary
network by Datapoint Corporation in the
late 1970s. ARCnet became very popular
when Standard Microsystems Corporation
(SMC) developed a chip set for PCs. The
architecture has been used for years and has
become a de facto standard. However, it has
not become as popular as other network
architectures, such as Ethernet. ARCnet is
popular for smaller networks because it is
relatively simple to set up and operate, its
components are inexpensive (street prices
for ARCnet boards are among the lowest),
and the architecture is widely supported.
ARCnet has a transmission rate of 2.5
megabytes per second (Mbps). ARCnet Plus
is a newer, 20 Mbps version. A third-party,
100 Mbps architecture based on ARCnet
is also available from Thomas-Conrad.
Although ARCnet Plus was developed by
Datapoint Corporation alone, current and
future development of ARCnet standards is
under the aegis of the ATA (ARCnet Trade
Association), a consortium of vendors that
market ARCnet products.
ARCnet uses token passing to control
access to the network. Each node in an
ARCnet network has a unique address
(between 1 and 255), and the token is
passed sequentially from one address to the
next. Nodes with successive addresses are
not necessarily next to each other in the
physical layout.
Officially, ARCnet uses a bus topology,
but in practice ARCnet networks can use a
star or a bus wiring scheme. These two types
of networks use slightly different compo-
nents and are sometimes referred to as low-
impedance and high-impedance ARCnet,
respectively.


Prev Page 56 Next

44
ARCnet (Attached Resource Computer Network)
The figure "Context and properties of
ARCnet" summarizes the characteristics
of this architecture.
CONTEXT AND PROPER TIES
OF ARCNET
Context
Network Architecture
Shared-Media

ARCnet

Ethernet

Token Ring
Switched Media
Description
Shared-media, baseband network
Topology
Bus (high-impedance ARCnet)

Star (low-impedance ARCnet)
Access method
Token passing
Speed
Up to 2.5 Mbps
Cable
RG-62 coaxial (93-ohm)

Unshielded twisted-pair

Fiber-optic
Frame size
Up to 508 data bytes
Variants
High-impedance ARCnet

Low-impedance ARCnet

Mixed-impedance ARCnet

ARCnet Plus
ARCnet
The hardware components needed in an
ARCnet network include an ARCnet net-
work interface card, cable, connectors, hubs,
active links, and baluns.
ARCnet NICs include chips to handle
the ARCnet protocols and packet formats,
as well as a transceiver (usually with a BNC
connector) on the card. Most ARCnet NICs
have a low-impedance transceiver, which is
best suited for a star or tree topology. (A tree
topology has features of both star and bus
topologies.) Cards with high-impedance
transceivers are suitable for a bus topology.
ARCnet cards do not come with hard-
ware addresses in a ROM chip. Instead, they
have jumpers that can be set to specify an
address for the node in which the card is
installed. The network administrator needs
to set this address (which must be between 1
and 255) for each card in the network. Each
node must have a unique address. The net-
work administrator also needs to set the
IRQ (interrupt) and I/O (input/output)
addresses on the card. The hardware address
is network-dependent; the IRQ and I/O
addresses are machine-dependent.
ARCnet cable can be coaxial, twisted-pair,
or even fiber-optic. Coaxial ARCnet net-
works generally have RG-62 cable, which
ARCnet Network Components
ARCnet Network Interface Card (NIC)
Cable


Prev Page 57 Next

ARCnet (Attached Resource Computer Network)
45
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
has a 93-ohm impedance. Other types of
coaxial cable, such as RG-59U or RG-11U,
are also used.
An ARCnet network might include
unshielded twisted-pair (UTP) or IBM's
special-design cables (Types 1 and 3), but
only if the NIC has the appropriate connec-
tors or if an appropriate adapter is available.
If UTP cabling is used, nodes are arranged
in a daisy chain and one end of the chain
is connected to a hub or to an adapter that
connects to coaxial cable. Similar converters
can convert from coaxial to fiber-optic
cable.
The last node in an ARCnet network
must be terminated with a resistor of appro-
priate strength: 93 ohm for coaxial net-
works and 105 ohm for networks using
twisted-pair wiring.
For coaxial cable, BNC connectors are
used. For twisted-pair cable, the connectors
are either the modular RJ-11/RJ-45 tele-
phone type, or the D-shell type used for
standard serial and parallel ports.
Active links are boxes used to connect
two cable segments when both cable
segments have high-impedance NICs
connected.
Baluns are used to connect coaxial and
twisted-pair cabling.
Connectors, Active Links, and Baluns
Hubs serve as wiring concentrators. Three
types of hubs can be used:
Active hubs: Active hubs have their own
power supply. They can clean and
boost a signal and then relay it along
the network. An active hub serves as
both a repeater and a wiring center.
Active hubs usually have 8 ports, but
they can have as many as 64. The type
of hub used must be appropriate for
the type of cable being used. Active
hubs can extend the maximum dis-
tance between nodes.
Passive hubs: Passive hubs simply relay
signals without cleaning or boosting
them. These types of hubs collect wir-
ing from nodes and must be connected
to an active hub. Passive hubs have
four ports and are used only in low-
impedance networks. Passive hubs
cannot be used to extend the distance
between nodes.
Intelligent hubs: Intelligent hubs are
active hubs that use a low-frequency
signal band to monitor the status of
a link. These hubs can have up to
16 ports.
ARCnet data transmissions are broadcast to
all nodes on the network (a feature charac-
teristic of both bus and star topologies), but
Hubs
ARCnet Operation


Prev Page 58 Next

46
ARCnet (Attached Resource Computer Network)
the transmitted packets are (presumably)
read only by the node(s) to which the desti-
nation address applies. Note that even
though all nodes can listen at the same time,
only one node can transmit.
ARCnet has several different types of
frames, or packets, which are listed on the
table "ARCnet Packets." The figure "ARC-
net frame structure" shows the makeup of
ARCnet frames.
The data, control, or check bytes that
make up the frame are known as ISUs
(information symbol units). ISUs are defined
differently in ARCnet and in ARCnet Plus.
All ARCnet frames begin with a 6-bit
alert signal, and all bytes begin with the
Structure of an ARCnet Packet
ARCNET PACKETS
PACKET TYPE
FUNCTION
ITT (Invitation to
Transmit)
FBE (Free Buffer
Enquiry)
ACK (Acknowledge)
NAK (Negative
Acknowledge)
PAC
The token, which deter-
mines the node that is
allowed to transmit
The frame that is used to
ask whether the destina-
tion node is able to receive
packets
The packet used to indi-
cate that a packet was
received as transmitted
The packet used to indi-
cate that a packet was not
received correctly and
should be retransmitted
The actual ARCnet data
frame
bit sequence 110, so that each byte actually
requires 11 bits in an ARCnet transmission.
ARCnet data frames consist of data,
header, and trailer. Originally, an ARCnet
frame could have up to 252 bytes of data.
Almost all ARCnet implementations now
support an expanded frame of up to 508
bytes of data (plus a dozen or so header
bytes).
An ARCnet header for a PAC frame
includes the following:
I A start of header byte
I Source and destination addresses,
with values between 1 and 255 (a des-
tination address of 0 indicates that the
frame is being broadcast to all nodes)
I One or two bytes indicating the num-
ber of data bytes
The trailer is a 16-bit CRC (cyclic redun-
dancy check) value.
The transmission of data frames in an ARC-
net network is controlled by a token, which
is a special data frame. This token, in turn, is
dispensed by the network's controller, which
is the node with the lowest address. The
controller is determined when the network
is first activated. Each node broadcasts
its address and the node with the lowest
address becomes the controller. This recon-
figuration process, which takes less than a
tenth of a second, is repeated each time a
new node joins the network.
The controller passes the token sequen-
tially from one address to the next. The
node with the token is the only node
allowed to transmit, with some exceptions.
Data Frame Transmission


Prev Page 59 Next

ARCnet (Attached Resource Computer Network)
47
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
ARCNET FRAME STR UCTURE
SD ENQ DID DID
SD
Frame components are symbols containing
the following:
SD
Starting delimiter, a special bit

pattern of six consecutive 1bits, to

indicate the start of the frame
EQT
ASCII 0x04, which indicates the

frame type
NID
The address of the next node to

get the token
ENQ
ASCII 0x85, which identifies the

frame type
DID
The address of the destination node

for the enquiry
ACK
ASCII 0x86, indicating that the

packet was recieved correctly
NAK
ASCII 0x15, indicating that the

packet was not recieved correctly
SOH
ASCII 0x01, indicating the start of

the header
SID
The address of the source node

sending the frame
CP
A continuation pointer value,

indicating the number of data bytes
SC
System code
DATA
Up to 508 symbols containing

system code and data
FCS
Frame check sequence, verifying the

integrity of the frame
Starting Delimiter (SD)
ITT Frame
FBE Frame
SD SOH SID DID DID CP SC DATA FSC
PAC Frame
ACK
ACK Frame
NAK Frame
1 or 2 1 or 2 0-508
2
1
1
1
1
1
SD EQT NID NID
SD NAK


Prev Page 60 Next

48
ARCnet (Attached Resource Computer Network)
Frame transmission is a complicated pro-
cess in ARCnet. A node (the source) waiting
to send a message to another node (the desti-
nation) needs to do several things, in the fol-
lowing order:
1. The source waits for the token (ITT).
2. Once it has the token, the source sends
an FBE packet to the destination to
make sure the destination has room for
the frame.
3. The source waits for a positive reply.
4. Once the source gets a positive
response (ACK) to the FBE packet,
the source broadcasts the frame.
5. The source waits for an acknowl-
edgment from the intended desti-
nation. The destination node must
acknowledge receipt of the frame.
Since acknowledgment is required,
ARCnet can guarantee frame delivery.
6. Once the frame has been received
at the destination, the controller passes High-Impedance ARCnet
the token to the next address.
Unless something is wrong on the network,
every node gets the token at least once every
840 milliseconds. If a node has not seen the
token within that time, that node can dis-
rupt the network and force the creation of
a new token by sending a reconfiguration
burst-a predefined bit pattern sent hun-
dreds of times in succession-to destroy the
existing token. After a period, the token is
regenerated, the network nodes reannounce
Disrupting Data Transmission
themselves, and the network begins trans-
mitting again.
New nodes on an ARCnet network also
send a reconfiguration burst. This pattern
announces their presence on the network,
and possibly establishes a new node as
controller.
ARCnet's small frame size causes compati-
bility problems with some network-layer
protocols, such as Novell's IPX protocol.
IPX passes 576-byte packets (known as
datagrams) to the architecture operating at
the data-link layer. This packet size is too
large, even for an extended ARCnet frame.
To enable IPX to talk to ARCnet, the
fragmentation layer was developed. At this
layer, the source node breaks an IPX packet
into two smaller frames for ARCnet. At the
destination's fragmentation layer, the data-
gram is reassembled before being passed
to IPX.
High-impedance ARCnet networks use a
bus topology, as illustrated in the figure
"Layout for a high-impedance ARCnet
network." The high-impedance NICs make
it possible to daisy chain nodes and active
hubs. The active hubs serve as collectors for
other hubs and nodes.
The following restrictions apply to high-
impedance ARCnet networks:
I No single cable segment connecting
nodes can be be more than 305 meters
(1,000 feet) long.
Communicating with Higher Layers


Prev Page 61 Next

ARCnet (Attached Resource Computer Network)
49
a
b
c
d
efghijklmnopqrstuvwxyz
LAYOUT FOR A HIGH-IMPEDANCE ARCNET NETWORK


Prev Page 62 Next

50
ARCnet (Attached Resource Computer Network)
I Only active (or intelligent) hubs may
be used.
I Adjacent active hubs (hubs with no
intervening nodes) must be within 610
meters (2,000 feet).
I Nodes are connected to the trunk cable
using BNC T-connectors. The node's
NIC must be connected directly to the
T-connector; that is, drop cable is not
allowed.
I T-connectors must be at least 1 meter
(3.25 feet) apart on the cable.
I At most, eight nodes can be connected
in a series (with no intervening hubs).
I Both ends of a cable segment must be
terminated with either a BNC termina-
tor or an active hub (or link).
I The cabling cannot loop back on itself.
For example, the cable cannot go from
an active hub through other hubs and
eventually connect back into the origi-
nal hub.
Low-impedance ARCnet networks use a star
topology, in which passive hubs serve to col-
lect nodes, as illustrated in the figure "Lay-
out for a low-impedance ARCnet network."
Each passive hub is connected to an active
hub. Active hubs can be linked with each
other, and they can also be linked directly
Low-Impedance ARCnet
with nodes. In the latter case, the active hub
also acts as a wiring center.
The following restrictions apply to low-
impedance ARCnet networks:
I Active hubs can be connected to
nodes, active hubs, or passive hubs.
The active hub must be within 610
meters (2,000 feet) of an active hub or
a node, or within 30 meters (100 feet)
of a passive hub.
I Passive hubs can be used only between
a node and an active hub; two passive
hubs cannot be next to each other. A
passive hub must be within 30 meters
(100 feet) of an active hub and within
30 meters (100 feet) of a node.
I Nodes can be attached anywhere on
the network, provided the node is
within the required distance of an
active or passive hub: within 610
meters (2,000 feet) of an active hub
or within 30 meters (100 feet) of a
passive hub.
I Unused hub ports must be terminated
on a passive hub and should be termi-
nated on an active hub.
I The cabling cannot loop back on itself.
For example, the cable cannot go from
an active hub through other hubs and
eventually connect back into the origi-
nal hub.


Prev Page 63 Next

ARCnet (Attached Resource Computer Network)
51
a
b
c
d
efghijklmnopqrstuvwxyz
LAYOUT FOR A LOW-IMPEDANCE ARCNET NETWORK


Prev Page 64 Next

52
ARCnet (Attached Resource Computer Network)
A mixed ARCnet network is one that
includes both high- and low-impedance
components in the same network, as
illustrated in the figure "Layout of a
mixed-impedance ARCnet network, with
low-impedance and high-impedance com-
ponents." In this type of network, all the
restrictions for both impedance levels
must be observed.
Perhaps the most important constraint
for a mixed-impedance ARCnet is that high-
impedance NICs can be used in place of
low-impedance cards, but the reverse is not
possible. Because of this restriction, it is cru-
cial that you keep track of what kind of NIC
is in each node.
The following restrictions apply to both
high- and low-impedance ARCnet networks:
I The maximum length of a cable seg-
ment depends on the type of cable. The
general restriction is that the signal
attenuation must be less than 11 dB
over the entire cable segment at a fre-
quency of 5 MHz. In practice, this
leads to the following maximum
distances:
Coaxial cable: 450600 meters (1,500
2,000 feet)
UTP and IBM Type 3 (unshielded) cable:
100 meters (330 feet)
Mixed-Impedance ARCnet
Restrictions on ARCnet Networks
IBM Type 1 (shielded) cable:
200 meters (660 feet)
I The maximum cable length for the
entire network is 6,000 meters (20,000
feet)
I The maximum number of cable seg-
ments in a series is three. If UTP cable
is used, the series of segments can be at
most about 130 meters (430 feet); for
coaxial cable, the maximum length is
about 300 meters (990 feet).
I Each cable segment must be termi-
nated at both ends by being connected
to an active hub or terminator.
I An ARCnet network can have a maxi-
mum of 255 nodes. Each active hub
counts as a node.
I At most, 10 nodes are allowed in a
series when UTP cable is used; 8 nodes
if coaxial cable is used.
I The maximum distance between any
two nodes on the network is deter-
mined by the constraint that no ARC-
net signal can have a propagation
delay of more than 31 microseconds.
The total propagation delay is deter-
mined by adding the propagation
delays in all the devices (nodes, hubs,
and cable) connecting the nodes. Net-
work components generally have
propagation delays of less than 0.5
microseconds, and much less in some
cases.


Prev Page 65 Next

ARCnet (Attached Resource Computer Network)
53
a
b
c
d
efghijklmnopqrstuvwxyz
LAYOUT OF A MIXED-IMPEDANCE ARCNET NETWORK,
WITH LOW-IMPEDANCE AND HIGH-IMPEDANCE COMPONENTS


Prev Page 66 Next

54
ARCnet (Attached Resource Computer Network)
ARCnet has the following advantages:
I Components are relatively inexpen-
sive. Street prices for basic ARCnet
NICs usually are less than those for
Ethernet or Token Ring NICs.
I Because the ARCnet architecture and
the chip set have been around a long
time, the hardware has become stable.
The result is that there are few compat-
ibility or reliability problems with
ARCnet components.
I Wiring is very flexible, allowing for
lots of leeway in placing nodes.
I It is relatively easy to use different
types of cabling in an ARCnet network
(but adapters must be used to avoid
connection incompatibilities).
I A star layout makes diagnostics easy in
low-impedance networks.
I Except for the extra cabling a star
topology requires, installation is rela-
tively inexpensive.
ARCnet has the following disadvantages:
I Its data transmission is inefficient.
ARCnet sends three overhead bits
for every byte. Also, administrative
exchanges (such as ACK or NAK
packets) between source and destina-
tion are done on the data bandwidth,
which degrades performance further.
I Actual throughput is much less than
the maximum 2.5 Mbps. Even for
small networks, the throughput is less
than 65 percent of maximum, and this
value decreases as more nodes are
added to the network.
I The network administrator must man-
ually set a unique address by adjusting
switches on every NIC in the network.
If two nodes have the same address,
the administrator will need to track
down the conflicting boards by tedious
examination of each NIC.
I Because of throughput and addressing
restrictions, ARCnet is not particularly
well-suited for internetworking.
Datapoint recently released ARCnet Plus, a
20-Mbps version of the ARCnet standard.
ARCnet Plus has the following features:
I Backward-compatibility with ARCnet
I Ability to communicate with both
ARCnet and ARCnet Plus nodes
I Support for transmission rates of up to
20 Mbps
I
Support for data frames up to 4,224
bytes long
I Use of the same RG-62 cable as ordi-
nary ARCnet
I New frames, with enhanced frame
formats and command sets
I Support for up to 1 MB of buffer space
ARCnet Plus achieves its greater speed by
cutting the time interval for a symbol in half
and by using phase and amplitude shifting to
encode four bits in every signal; that is, the
ARCnet Advantages
ARCnet Disadvantages
ARCnet Plus


Prev Page 67 Next

ARPAnet (Advanced Research Projects Agency Network)
55
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
basic symbol in ARCnet Plus is actually a
nibble.
Like its predecessor, ARCnet Plus regu-
lates much network activity by timing. The
allowable intervals are much smaller with
ARCnet Plus, however. For example, a bit
interval is half as long in ARCnet Plus as
in regular ARCnet.
Another extension of this type of
architecture is TCNS, offered by Thomas-
Conrad, which is a 100 Mbps, copper-
based network.
BROADER CATEGOR Y
Network Architecture
SEE ALSO
TCNS (Thomas-Conrad Network
System)
TIPS ON ARCNET ADDRESSES
Keep accurate addresses. Make sure you have up-
to-date records of the address set for each ARC-
net node's NIC. When you need to find duplicate
addresses or add nodes, you'll be glad you did.
If you're the administrator, never let anyone
else change the node addresses, because you may
have to deal with the problems caused by their
sloppiness.
Assigning the low address is particularly impor-
tant. The network controller will be the node
with the lowest address, so make sure this
machine is fast enough to handle the controlling
role. In general, it's best to assign the lowest
addresses to servers, bridges, and routers.
MARF (Alarm Reporting Function)
In the OSI network management model,
a service that reports failures, faults, or
problems that might become faults.
M
ARM (Asynchronous Response Mode)
In the ISO's HLDC (High-Level Data Link
Control) protocol, ARM is a communica-
tions mode in which a secondary (slave)
node can initiate communications with a
primary (master) node without first getting
permission from the primary node.
ARM's operation is in contrast to NRM
(normal response mode), in which the pri-
mary node must initiate any communica-
tion, and to ABM (asynchronous balanced
mode), in which the two nodes are equal.
BROADER CATEGOR Y
HDLC (High-Level Data Link Control)
M
ARPA (Advanced Research Projects
Agency)
The agency that was largely responsible for
what eventually became the Internet. Now
called DARPA (for Defense ARPA).
M
ARPAnet (Advanced Research Projects
Agency Network)
ARPAnet was the first large-scale, packet-
switched, wide-area network (WAN). It was
originally developed in the early 1970s
under the auspices of the U.S. Department
of Defense's Defense Advanced Research
Projects Agency (DARPA).


Prev Page 68 Next

56
ARQ (Automatic Repeat Request)
Many of the most commonly used net-
working protocols, including TCP/IP, were
developed as part of the ARPAnet project.
The ARPAnet was decommissioned in 1991,
but parts of the network have become part
of the Internet.
M
ARQ (Automatic Repeat Request)
In communications, a control code that
indicates an error in transmission and
that requests a retransmission.
M
ARS (Automatic Route Selection)
In telephony, a process by which a path
is selected for a transmission; also called
alternate route selection.
MAS (Autonomous System)
In the Internet world, AS (autonomous sys-
tem) is a term for a collection of routers that
are part of a larger network but that are
under the control of a single organization.
The routers, or gateways as they are called
in the older Internet terminology, communi-
cate with each other using a common proto-
col, known as an interior gateway protocol
(IGP). Currently, the two most widely sup-
ported IGPs in the Internet community are
the OSPF (Open Shortest Path First) and the
Integrated IS-IS protocols.
ASs communicate using an exterior
gateway protocol, such as EGP (Exterior
Gateway Protocol) and BGP (Border Gate-
way Protocol).
In the OSI Reference Model, an autono-
mous system is known as a routing domain,
IGPs are known as intradomain routing pro-
tocols, and EGPs are known as interdomain
routing protocols.
M
AS/400
A minicomputer line from IBM. The AS/400
was introduced in 1988 to replace the Sys-
tem/36 and System/38 series.
M
ASCII (American Standard Code for
Information Interchange)
ASCII is the character-encoding system used
most commonly in local-area networks
(LANs). The standard ASCII characters are
encoded in seven bits and have values
between 0 and 127. The remaining 128
characters form the extended ASCII charac-
ter set, whose elements may be defined dif-
ferently depending on the language being
used. See the tables "Standard ASCII Char-
acter Set" and "Extended ASCII Character
Set (IBM PC)."
In common usage, ASCII is used to refer
to a text-only file that does not include spe-
cial formatting codes.
BROADER CATEGOR Y
Encoding
COMPARE
EBCDIC


Prev Page 69 Next

ASCII (American Standard Code for Information Interchange)
57
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
STANDARD ASCII CHARACTER SET
DECIMAL
CHARACTER
DECIMAL
CHARACTER
DECIMAL
CHARACTER
0
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
NUL (null)
SOH (start
of heading)
STX (start of text)
ETX (end of text)
EOT (end of
transmission)
ENQ (enquire)
ACK (acknowledge)
BEL (bell)
BS (backspace)
HT (horizontal tab)
LF (line feed)
VT (vertical tab)
FF (form feed)
CR (carriage return)
SO (shift out)
SI (shift in)
DLE (data link
escape)
DC1 (device
control 1)
DC2 (device
control 2)
DC3 (device
control 3)
DC4 (device
control 4)
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
NAK (negative
acknowledge)
SYN (synchronous
idle)
ETB (end trans-
mission block)
CAN (cancel)
EM (end of medium)
SUB (substitute)
ESC (escape)
FS (file separator)
GS (group
separator)
RS (record
separator)
US (unit separator)
space
!
"
#
$
%
&
' (apostrophe)
(
)
*
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
+
, (comma)
-
.
/
0
1
2
3
4
5
6
7
8
9
:
;
<
=
>
?
@
A
B
C


Prev Page 70 Next

58
ASCII (American Standard Code for Information Interchange)
DECIMAL
CHARACTER
DECIMAL
CHARACTER
DECIMAL
CHARACTER
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
D
E
F
G
H
I
J
K
L
M
N
O
P
Q
R
S
T
U
V
W
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
X
Y
Z
[
\
]
^
_

a
b
c
d
e
f
g
h
i
j
k
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
{
|
}
~
DEL


Prev Page 71 Next

ASCII (American Standard Code for Information Interchange)
59
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
EXTENDED ASCII CHARACTER SET
DECIMAL
CHARACTER
DECIMAL
CHARACTER
DECIMAL
CHARACTER
0
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
b
a
c
d
f
e
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
y
z
1
2
3
4
!
"
#
$
%
&
'
(
)
*
+
,
_
.
/
1
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
2
3
4
5
6
7
8
9
:
;
<
=
>
?
@
A
B
C
D
E
F
G
H
I
J


Prev Page 72 Next

60
ASCII (American Standard Code for Information Interchange)
DECIMAL
CHARACTER
DECIMAL
CHARACTER
DECIMAL
CHARACTER
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
K
L
M
N
O
P
Q
R
S
T
U
V
W
X
Y
Z
[
\
]

_
`
a
b
c
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
{
|
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
}


























Prev Page 73 Next

ASCII (American Standard Code for Information Interchange)
61
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
DECIMAL
CHARACTER
DECIMAL
CHARACTER
DECIMAL
CHARACTER
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174

















175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199

200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224


Prev Page 74 Next

62
ASCIIbetical Sorting
M
ASCIIbetical Sorting
A sorting strategy that uses the ASCII
character set as the basis for the ordering.
In ASCII, numbers and special symbols pre-
cede letters; uppercase letters precede lower-
case ones.
M
ASE (Application Service Element)
In the OSI Reference Model, an ASE (appli-
cation service element) is any of several ele-
ments that provide the communications and
other services at the application layer. An
application process (AP) or application
entity (AE) requests these services through
predefined interfaces, such as those provided
by APIs (Application Program Interfaces).
ASEs are grouped into common applica-
tion service elements (CASEs) and specific
application service elements (SASEs). The
CASEs provide services for many types of
applications; the SASEs represent or provide
services for specific applications or genres.
The following CASEs are commonly used:
ACSE (Association Control Service
Element): This element establishes
the appropriate relationship between
two applications (AEs) to enable the
applications to cooperate and commu-
nicate on a task. Since all associations
or relationships must be established
through the ACSE, and since applica-
tions must establish a relationship to
communicate, the ACSE is needed by
all applications.
CCRSE (Commitment, Concurrency, and
Recovery Service Element): This ele-
ment is used to implement distributed
CASE
DECIMAL
CHARACTER
DECIMAL
CHARACTER
DECIMAL
CHARACTER
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255


Prev Page 75 Next

ASI (Adapter Support Interface)
63
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
transactions which may require multi-
ple applications. The CCRSE helps
ensure that distributed data remains
consistent by making sure that applica-
tions do not interfere with each other
when doing their work and that
actions are performed completely or
not at all.
ROSE (Remote Operations Service Ele-
ment): This element supports interac-
tive cooperation between two appli-
cations, such as between a client and
a server. ROSE provides the services
needed for the reliable execution of
requested operations and transfer
of data.
RTSE (Reliable Transfer Service Ele-
ment): This element helps ensure that
PDUs (protocol data units), or packets,
are transferred reliably between appli-
cations. RTSE services can sometimes
survive an equipment failure because
they use transport-layer services.
The following SASEs are commonly used:
DS (Directory Service): This element
makes it possible to use a global direc-
tory, which is a distributed database
with information about all accessible
network entities in a communications
system.
FTAM: (File Transfer Access and Man-
agement): This element enables an
application to read, write, or otherwise
manage files on a remote machine.
JTM (Job Transfer and Manipulation):
This element enables an application to
SASE
do batch data processing on a remote
machine. With JTM, a node could, for
example, start a computation on a
supercomputer at a remote location
and retrieve the results when the com-
putation was complete.
MHS (Message Handling System):
This element enables applications to
exchange messages; for example, when
using electronic mail.
MMS (Manufacturing Message Service):
This element enables an application on
a control computer to communicate
with an application on a slave machine
in a production line or other auto-
mated operation.
VT (Virtual Terminal): This element
makes it possible to emulate the behav-
ior of a particular terminal, which
enables an application to communicate
with a remote system without consid-
ering the type of hardware sending or
receiving the communications.
The entire set of ASEs required for a par-
ticular application is known as the applica-
tion context (AC) for that application.
BROADER CATEGOR Y
AC (Application Context)
M
ASI (Adapter Support Interface)
ASI (Adapter Support Interface) is a
standard interface developed by IBM for
enabling Token Ring adapters to talk to any
of several higher-level protocols. The most
recent version of ASI is marketed as LAN
Support Program.


Prev Page 76 Next

64
Asia and Oceania Workshop (AOW)
Like other adapter interfaces, such
as NDIS (Network Driver Interface Speci-
fication) by Microsoft and ODI (Open
Data-Link Interface) by Novell, ASI includes
at least the following two components:
I A data-link-layer driver to talk to the
network interface card (NIC)
I A network-layer driver to talk to the
network-level protocols
M
Asia and Oceania Workshop (AOW)
SEE
AOW (Asia and Oceania Workshop)
M
ASIC (Application-Specific Integrated
Circuit)
Special-purpose chips with logic designed
specifically for a particular application or
device. ASICs are also known as gate arrays,
and they are constructed from standard cir-
cuit cells from a library.
M
ASN.1 (Abstract Syntax Notation
One)
In the OSI Reference Model, ASN.1
(Abstract Syntax Notation One) is a nota-
tion used to describe data structures, such as
managed objects in a network management
system.
ASN.1 is machine-independent and is
used in many networking contexts. For
example, it is used to describe application-
layer packets in both the OSI network
management framework and in the Simple
Network Management Protocol (SNMP)
from the Internet TCP/IP protocol suite.
ASN.1 serves as a common syntax for
transferring information between two end
systems (ESs) that may use different encod-
ing systems at each end.
PRIMAR Y SOURCES
CCITT recommendations X.208 and
X.209; ISO documents 8824 and 8825
BROADER CATEGOR Y
Abstract Syntax
SEE ALSO
BER (Basic Encoding Rules)
M
Asserted Circuit
A circuit that is closed; that is, a circuit with
a voltage value. Depending on the logic
being used, an asserted circuit can represent
a 1 (usually) or 0 (rarely).
M
Assigned Number
In the Internet community, a numerical
value that serves to distinguish a particular
protocol, application, or organization in
some context. For example, assigned num-
bers distinguish the different flavors of
Ethernet protocols used by different imple-
menters. Assigned numbers, which are not
addresses, are assigned by the Internet
Assigned Numbers Authority (IANA).
M
ASVD (Analog Simultaneous
Voice/Data)
A proposed modem standard that can be
used to transmit multimedia materials-
voice, video, etc.-over ordinary (analog)
telephone lines. The ASVD specifications are
being finalized by the ITU (International


Prev Page 77 Next

AT Command Set
65
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
Telecommunication Union, formerly known
as the CCITT).
ASVD is offered as an inexpensive (and
slower) alternative to ISDN (Integrated Ser-
vices Digital Network). The bandwidth for
ASVD is considerably more limited than for
ISDN. The version under consideration sup-
ports modem speeds of up to 14.4 kbps, but
somewhat slower speeds for multimedia
data.
M
Asynchronous
Asynchronous describes a communications
strategy that uses start and stop bits to indi-
cate the beginning and end of a character,
rather than using constant timing to trans-
mit a series of characters. In a sense, asyn-
chronous transmissions actually synchronize
for each character. The figure "A data word
sent by asynchronous transmission" shows
the bits used in this communications
method.
Asynchronous communications methods
are generally less efficient but more resistant
to disruption than synchronous communica-
tions. Asynchronous methods are more effi-
cient for situations in which traffic comes in
bursts (rather than moving at a regular
pace). Common examples of asynchronous
communications devices are modems and
terminals.
M
Asynchronous Modem Eliminator
(AME)
SEE
AME (Asynchronous Modem Eliminator)
MATA (ARCnet Trade Association)
A consortium of vendors and other organi-
zations that manages ARCnet specifications.
M
AT Command Set
The AT command set was developed by
Hayes Microcomputer Products to operate
its modems. The AT in the title is an abbre-
viation for attention. This signal precedes
most of the commands used to get a modem
to do its work. For example, ATDP and
ATDT (for attention dial pulse and attention
dial tone, respectively) are used to dial a
number on either a pulse or Touch Tone
phone.
The AT command set quickly became a
de facto standard. It is now used by most
modem manufacturers and is supported on
virtually every modem on the market.
SEE ALSO
Modem
A DATA WORD SENT BY
ASYNCHRONOUS TRANSMISSION


Prev Page 78 Next

66
ATCON
M
ATCON
A Novell NetWare program that monitors
the AppleTalk protocol stack in a multipro-
tocol network. It reports statistics about
the performance of AppleTalk devices
and services.
M
ATDP (Attention Dial Pulse)
In the Hayes modem command set, a com-
mand to dial a number using a pulse (rotary)
telephone.
SEE ALSO
AT Command Set
M
ATDT (Attention Dial Tone)
In the Hayes modem command set, a com-
mand to dial a number using a Touch Tone
phone.
SEE ALSO
AT Command Set
MATM (Asynchronous Transfer Mode)
ATM (Asynchronous Transfer Mode) is a
packet-switched, broadband network archi-
tecture that is expected to become an estab-
lished standard by the late 1990s. It forms
the core of a broadband ISDN (BISDN)
architecture, which extends the digital trans-
mission capabilities defined by ISDN to
allow data, voice, and multimedia transmis-
sions on the same lines. It is also known as
cell relay, to distinguish it from frame relay.
ATM is a real-time architecture that will
be able to provide very high bandwidths as
needed. The initial implementations will
operate at 155.52 megabits per second
(Mbps), then at 622.08 Mbps. Speeds up
to 2.488 gigabits per second (Gbps) are
planned and have been demonstrated in
limited tests.
The very high bandwidth and the ability
to transmit multiple media make ATM an
attractive, high-speed architecture for both
local-area networks (LANs) and wide-area
networks (WANs). It is useful for enterprise
networks, which often connect LANs over
wide areas and may need to transport large
amounts of data over very long distances.
Long-haul, high-bandwidth capabilities
are particularly attractive for WANs, which
have until now been shackled by the rela-
tively low bandwidths over long-distance
lines. FDDI (Fiber Distributed Data Inter-
face) is a good architecture for LANs, and
frame relay has possibilities for WANs, but
neither of these architectures is suitable for
both LANs and WANs. But note that ATM
is still quite expensive.
The figure "Context and properties of
ATM" summarizes the characteristics of this
architecture.
ATM has the following features:
I Transmission over fiber-optic lines.
These can be local or long-distance,
public or private lines. Long-distance
lines can be leased or dial-up.
I Capability for parallel transmissions,
because ATM is a switching architec-
ture. In fact, each node can have a ded-
icated connection to any other node.
ATM Features


Prev Page 79 Next

ATM (Asynchronous Transfer Mode)
67
a
b
c
d
e
f
g
hi
j
k
l
m
n
o
p
q
rs
t
u
v
w
x
y
z
I Operation at maximum speed at all
times, provided there is enough net-
work traffic to give the required
throughput.
I Use of fixed-length (53-byte) packets,
which are known as cells.
I Error correction and routing in hard-
ware, partly because of the fixed cell
sizes.
I Transmission of voice, video, and data
at the same time. The fixed-length cells
also make voice transmission more
CONTEXT AND PROPER TIES OF ATM
Context
Network Architecture
Shared-Media
Switched-Media

Circuit

Message

Packet

Fixed-Size

ATM (Cell Relay)

Variable-Size

Frame Relay
Broadband
Core of BISDN
Useful for LANs and WANs
Uses short- or long-haul fiber-optic
cable
Initial speeds up to 166.62 Mbps
(eventural speeds up to 2.49
Gbps)
Can always operate at top speed
(provided there is enough traffic)
Can transmit voice, video, data
(simultaneously, if necessary)
ATM
Properties
Structure
Layers
Planes
Cells
Physical
Users
Constant Size (53 Octets)
(Two Sublayers)
Management
(48-Octet Payload)
ATM Layer
Control
(5-Octet Header)
(Service Independent)
Not Byte-Bound/Oriented
AAL
(Two Sublayers)
(Four Service Classes)
A: for Voice, Data
B: for Video, etc.
C: for Connection-
Oriented Mode
D: for Connectionless
Mode


Prev Page 80 Next

68
ATM (Asynchronous Transfer Mode)
accurate, because there is less timing
variation.
I Easier load balancing, because the
switching capabilities make it possible
to have multiple virtual circuits
between sender and receiver.
The ATM architecture is organized into lay-
ers, as are other network architectures, and
also into planes, which specify domains of
activity. See the figure "Structure of the
ATM architecture" for a graphic representa-
tion of the organization of the planes and
layers.
The ATM physical layer corresponds to the
OSI Reference Model physical layer. It is
concerned with the physical medium and
interfaces, and with the framing protocols
(if any) for the network.
The physical layer has two sublayers.
The lower sublayer, physical medium (PM),
includes the definition for the medium (opti-
cal fiber) and the bit-timing capabilities. The
upper sublayer, transmission convergence
(TC), is responsible for making sure valid
cells are being created and transmitted. This
involves breaking off individual cells from
the data stream of the higher layer (the ATM
layer), checking the cell's header, and encod-
ing the bit values.
The user network interface (UNI) speci-
fied by the ATM forum, an organization
dedicated to defining and implementing
ATM, allows for various types of physical
ATM Structure
Physical Layer
interfaces for ATM networks, including the
following:
I SONET connections at 155.52 Mbps
(OC-3, STS-3, or in CCITT terminol-
ogy, STM-1)
I DS3 connections at 44.736 Mbps
I 100 Mbps connections using 4B/5B
encoding
STRUCTURE OF THE
ATM ARCHITECTURE


Prev Page 81 Next

ATM (Asynchronous Transfer Mode)
69
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
I 155 Mbps connections using 8B/10B
encoding
These interfaces all use optical fiber,
which is the only medium specified for
ATM. A work group is investigating the pos-
sibility of defining ATM for Category 3
unshielded twisted-pair (UTP) wire.
The ATM layer is a service-independent
layer at which cell headers and trailers are
created, virtual channels and paths are
defined and given unique identifiers, and
cells are multiplexed or demultiplexed. The
ATM layer creates the cells and uses the
physical layer to transmit them.
The topmost layer, AAL is service-
dependent. It provides the necessary
protocol translation between ATM and
the other communication services (such
as voice, video, or data) involved in a trans-
mission. For example, the AAL translates
between elements from a pulse-code modu-
lation (PCM) transmission (which encodes
voice data in digital form) and ATM cells.
The following four types of services,
which each use different AAL protocols,
are defined at the AAL:
I Class A is suited for constant bit
rate (CBR) data and provides circuit-
switching emulation. This is appropri-
ate for voice data. The protocol is
AAL 1.
I Class B is for variable bit rate (VBR)
data; for example, video transmissions
during teleconferences. The protocol is
AAL 2.
I Class C is suited for connection-
oriented data transmissions. The
protocol is AAL 3 or AAL 5.
I
Class D is suited for connectionless
data transmissions. The protocol is
AAL 4 or AAL 5.
AAL 5 supports classes C or D more
efficiently than AAL 3 or AAL 4.
The AAL has two sublayers:
I CS (convergence sublayer) is the upper
sublayer that provides the interface for
the various services. Users connect to
the CS through service access points
(SAPs). No protocol data units (PDUs)
are defined for this level because the
data passing through is application-
and service-dependent.
I SAR (segmentation and reassembly) is
the sublayer that packages variable-
size packets into fixed-size cells at the
transmitting end, and repackages the
cells at the receiving end. The SAR
sublayer is also responsible for finding
and dealing with cells that are out of
order or lost.
A separate PDU is defined for each
class of service. Each PDU contains 48
octets, which are allocated for the header,
trailer, and data (known as payload in ATM
terminology). Of these, the AAL 1 PDU can
carry the most data at a time: a 47-octet
payload. AAL 3 and AAL 4 each have a
ATM Layer
AAL (ATM Adaptation Layer)
AAL Sublayers


Prev Page 82 Next

70
ATM (Asynchronous Transfer Mode)
44-octet payload, and AAL 2 has a 45-octet
payload. These PDUs become the data (pay-
load) for the ATM cells that are transmitted.
Three domains of activity, known as planes,
are distinguished for ATM:
I The control plane, on which calls
and connections are established and
maintained.
I The user plane, on which users, or
nodes, exchange data. This is the plane
at which ordinary user services are
provided.
I The management plane, on which
network-management and layer-
management services are provided.
This plane coordinates the three planes
and manages resources for the layers.
The figure "ATM transmission elements"
shows the elements used as a transmission
gets onto an ATM network. The top part of
the illustration represents the higher (non-
ATM) service layers; the bottom part repre-
sents the ATM and physical layers in the
ATM model. The ATM node does the work
of the AAL and much of the ATM layer.
Data from the various types of services
(voice, video, data, and so forth) is handled
at the AAL layer in an ATM node. The data
is converted into ATM cells, regardless of
the types of packets that came in. The data
is handled by the appropriate class of ser-
vice. For example, the Class A services will
ATM Planes
ATM Operation
handle voice data; Class C or D services will
handle data from a network, and so forth.
Data comes into the AAL as packets of
varying sizes, but leaves as fixed-size (48-
octet) SAR PDUs. The details of these PDUs
depend on the type of service (Class A, B, C,
or D) being used. The SAR sublayer does the
necessary chopping and packing.
The SAR PDUs from the various services
are wrapped into ATM cells at the ATM
layer and multiplexed for transmission onto
the ATM cell stream. These ATM cells
contain the virtual channel and path iden-
tification required for the cell to reach its
destination. The ATM switch uses channel
and path information to send the cell out
through the appropriate port.
The cell stream contains bits and pieces of
various types of packets, all in separate cells.
The cells may be routed, or switched, at var-
ious points on their path, as appropriate for
maintaining connections at the required
quality of service.
The cell stream is encoded and trans-
mitted over the physical media connecting
the ATM network. At the receiving end, the
ATM routes the cells to the appropriate ser-
vices at the AAL. The cells are repackaged
into the appropriate packet form by the
AAL service. This service also checks that
the entire packet has been received and that
everything is correct.
At the receiving end, the transmission
sequence is undone, with the services at the
topmost (for ATM) sublayer unpacking the
ATM cells to reveal the various types of
data, which are passed out to the services
that handle the data.


Prev Page 83 Next

ATM (Asynchronous Transfer Mode)
71
a
b
c
d
efg
hi
j
kl
m
n
o
p
q
r
s
t
u
v
w
x
y
z
ATM TRANSMISSION ELEMENTS
Data Stream
(from network,
router)
Voice Stream
(from classic
telephone service)
Video Stream
(from teleconferences,
image transmissions)
ATM Cells
(from ATM stations)
Video Cell
Packet-Mode
Services
Circuit-Mode
Emulation
Video-Mode
Services
ATM
Services
Data Cell
Voice Cell
. . . .
Data Cell
Stream of ATM Cells
ATM Switch
(Multiplexer)
ATM Node
ATM Cells
ATM Cells
ATM Cells
ATM Cells


Prev Page 84 Next

72
ATM (Asynchronous Transfer Mode)
The ATM architecture distinguishes between
two interfaces:
I The user-network interface (UNI),
which connects an end-user to the
network via an ATM switch or other
device. This interface supplies network
access.
I The network-node interface (NNI),
which connects network nodes to each
other. This interface makes network
routing possible.
ATM cells are not byte oriented. Even
though cells are defined as a specific number
of octets, the fields within such a cell often
cross byte boundaries.
ATM cells consist of a five-octet header
and a 48-octet data, or payload, section. The
payload section is an SAR PDU, to which a
five-octet ATM header is added. See the fig-
ure "Structure of an ATM cell at the UNI."
Most of the bits in the header are used for
virtual path and channel identification. The
CLP (cell loss priority) bit indicates whether
the cell can be discarded if network traffic
volume makes this advisable. If the flag is
set, the cell is expendable.
Because header fields can extend over
multiple octets-for example, the VPI or
VCI fields-the ATM specifications include
the following guidelines for how bits are to
be arranged within a field:
I Within an octet, bit order goes from
left to right. For example, in octet
1, the VPI bits are-from highest to
lowest-bits 4, 3, 2, and 1, with
1 being the least significant bit within
that octet.
I Across octets, bit order goes down-
ward as octets go upward. Thus, the
lowest order bit in the VPI field is bit 5
in octet 2. Similarly, the lowest order
bit for the VCI field is bit 5 in octet 4;
the highest order bit in this field is bit
4 in octet 2, and the bits in octet 3
are between the high- and low-order
quartets.
The cell-structure shown in the
figure "Structure of an ATM cell at the
UNI" applies to cells that travel onto
the network across the UNI. When cells
ATM Interfaces
Cell Structure
STRUCTURE OF AN
ATM CELL AT THE UNI


Prev Page 85 Next

ATM (Asynchronous Transfer Mode)
73
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
are moving across the NNI-that is, for
routing purposes-the VPI field is extended
to encompass the entire first octet. That is,
cells at the NNI use 12 bits for VPI and 16
for VCI. There is no generic flow control
field for these cells.
Because ATM's progress toward becoming
the dominant high-speed architecture has
been much slower than anticipated, several
variants on the basic technology have been
proposed as a means of getting at least some
form of ATM into more markets and net-
works. Two of the more interesting variants
are ATM25 and ATM LAN emulation.
ATM25 is a 25 Mbps version proposed for
use in desktop networks-that is, in LANs.
This version was proposed by the Desktop
ATM25 Alliance, which includes IBM and
Apple among its members. This variant
would run on ordinary UTP (unshielded
twisted pair) cables, and would allow 25
Mbps transmissions in both directions. The
ATM Forum is considering the ATM25
specifications, and Alliance members are
currently working on specifications that
would enable products from different ven-
dors to work together, and that would
enable ATM25 networks to communicate
in a transparent manner with other, faster-
speed ATM networks.
This variant uses software to fool a network
operating system into thinking that an ATM
interface card is actually an Ethernet or
Token Ring adapter. This software may be
included as a driver on the workstation, or
client machine. Additional software runs a
LAN emulation server-either on an ATM
switch or on a separate PC.
With ATM LAN emulation, an ATM
device can be made to look like an Ethernet
or a Token Ring node to a network server.
Below the surface, however, the virtual
Ethernet device, for example, is able to oper-
ate at blazing ATM speeds by breaking the
Ethernet packets into ATM cells before
sending them on. The packets might be sent
across an ATM network to a receiving
device that also supports LAN emulation.
The packets could then be reassembled at
the receiving end and passed transparently
to a receiving Ethernet device. Information
in the header area identifies packets as com-
ing from a LAN emulation device. Such
an emulation makes ATM devices indepen-
dent of higher-level protocols (for example,
TCP/IP or IPX).
The ATM Forum is a consortium of several
hundred vendors, researchers, and other
involved parties. The Forum's charter is
to help develop and promote the use of
ATM-related products and services. Toward
this end, the forum provides information
about ATM, helps develop specifications
for ATM products and use, and generally
keeps ATM on the minds of the appropriate
people and groups.
Forum members are companies that are
interested in developing or using ATM tech-
nology. These companies are readying prod-
ucts for various facets of an ATM network,
such as nodes, switches, PBXs, and routers.
ATM Variants
ATM25
ATM LAN Emulation
ATM Resources


Prev Page 86 Next

74
ATPS (AppleTalk Print Services)
Various combinations of forum members/
vendors have formed partnerships to create
and market ATM components. Companies
such as Sprint and AT&T will offer ATM
services to their customers.
While many aspects of the ATM technol-
ogy and specifications are still in flux, signif-
icant portions have been tested and proven
viable. Vendors have forged ahead and are
selling ATM products. They are still quite
expensive, however, partly because the
absence of finalized specifications has led to
vendor-specific implementations. This, of
course, makes interoperability more elusive
and customers more reluctant.
ATM variants and emulation schemes
have been proposed in an effort to make
ATM better known. Major ATM vendors
have been cutting their prices, which is also
expected to help the established base grow.
BROADER CATEGORIES
Network Architecture; Network, Cell-
Switched; Network, Packet-Switched
M
ATPS (AppleTalk Print Services)
An NLM (NetWare Loadable Module) that
provides NetWare nodes with access to
printers and Macintosh nodes with access to
NetWare print queues. Settings for this mod-
ule are in the ATPS.CFG file.
M
ATTACH
In Novell's NetWare 2.x and 3.x, the
ATTACH command tells a file server that a
workstation exists and wants to join the net-
work. The server will assign the workstation
a connection number.
Once attached, the user at the worksta-
tion can access any of the server's services
(assuming that the user has the necessary
access rights to those services). The
ATTACH command cannot be used to
connect to the network initially. The LOGIN
command must be used for the first server.
Then the ATTACH command can be used to
attach to additional servers. ATTACH does
not execute a login script or redefine the
workstation's environment. The ATTACH
command is not included in NetWare 4.x.
BROADER CATEGOR Y
NetWare
M
Attachment
In electronic mail, an attachment is a file
that is sent along with a regular e-mail
message.
M
Attack Scanner
An attack scanner is a software package
used to probe UNIX networks for security
problems or flaws. The package will essen-
tially play the role of an intruder trying to
steal or force access to a network. The use of
such programs is somewhat controversial.
In April, 1995, a controversial attack
scanner product-SATAN (Security Analysis
Tool for Auditing Networks) by Wietse
Venema and Dan Farmer-was posted to
the Internet. Such a product can be used by
crackers (users trying to break into systems
for malicious purposes) as well as by system
administrators and security people. As a
result, the Internet community is divided as
to whether such a product should be made
freely available.


Prev Page 87 Next

Attribute
75
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
M
Attenuation
Attenuation is the loss of signal strength
over distance. It is measured in decibels (dB)
per kilometer (expressed as dB/km) or per
100 feet. In the logarithmic decibel scale, a 3
dB loss means a 50 percent loss in power, as
computed in the following equation. Specifi-
cally, the formula for power loss is:
In this equation, a 50 percent loss would
actually yield a result of -3 dB. Under cer-
tain conditions, the coefficient in the equa-
tion will be 20, in which case a result of -6
dB would indicate a 50 percent loss. When
describing losses, however, the negative sign
is dropped, so that a result of -6 dB is
expressed as a 6 dB loss.
Attenuation depends on several factors,
including the wire composition and size,
shielding, and frequency range of the signal.
For copper cable, attenuation increases with
signal frequency; for optical fiber, attenua-
tion is relatively constant over a large fre-
quency range.
Fiber-optic cable has the least attenua-
tion, usually fractions of a decibel per kilo-
meter. Unshielded untwisted-pair cable (such
as the silver, flat-satin cables used in short-
distance telephone and modem lines) has the
most attentuation of any cable types used in
telecommunications. This type of cable is
not used directly in networks.
M
Attenuation Factor
A value that expresses the amount of a
signal lost over a given distance, such as
decibel loss per kilometer (expressed as
dB/km).
M
Attribute
An attribute is a feature or property associ-
ated with an entity. For example, objects in
network management and entries in an
X.500 Directory Services database have
attributes.
An attribute has a type and a value asso-
ciated with it. The type constrains the form
the value can take. For example, an INTE-
GER type may have only a whole number
value, or a BOOLEAN may have only a
value that evaluates to TRUE or FALSE.
Much network management or monitor-
ing activity consists of determining or
changing attribute values. Attribute values
are read or set by functions that provide the
relevant network services.
Among the most important attributes are
those associated with files and directories,
because these ultimately limit what can be
done on a network. The attributes are gener-
ally represented as single-bit flag values,
with the flag either set or not set.
The specific attributes defined vary from
system to system, but attributes are used
in every operating system and networking
environment. Certain attributes assume or
replace others, and certain attributes over-
ride access rights. See the table "Novell Net-
Ware File and Directory Attributes" for
descriptions of NetWare attributes associ-
ated with files and directories.
dB = 10 log10 -----------------------
Powerout
Power
in
File and Directory Attributes


Prev Page 88 Next

76
Attribute
NOVELL NETWARE FILE AND DIRECTOR Y ATTRIBUTES
ATTRIBUTE
DESCRIPTION
A (Archive needed)
C (Copy inhibit)
Cc (Can't compress)
Co (Compressed)
Di (Delete inhibit)
Dc (Don't compress)
Dm (Don't migrate)
X (Execute only)
H (Hidden)
I (Indexed)
Ic (Immediate
compress)
M (Migrate)
P (Purge)
R (Rename inhibit)
Ra (Read audit)
Ro/Rw (Read only/
Read write)
S (Shareable)
Set automatically when a file is changed after its most recent backup. (NetWare 2.x,
3.x, 4.x)
Set to keep Macintosh files from being copied. Does not apply to DOS files. (NetWare
3.x, 4.x)
Set automatically when a file cannot be compressed because it would not save a signif-
icant amount of space. (NetWare 4.x)
Set automatically to show that a file has been compressed. (NetWare 4.x)
Set to keep users from deleting a file or directory. (NetWare 3.x, 4.x)
Set to prevent a file from being compressed. (NetWare 4.x)
Set to prevent a file from being migrated to a secondary storage medium, such as an
optical disk drive. (NetWare 4.x)
Set to keep a file from being copied, deleted, changed, or backed up. Since this setting
cannot be changed, it's necessary to keep a backup (nonrestricted) copy of the pro-
gram before freezing it. Assigning this attribute is not recommended; the same effect
can be accomplished with the Ro attribute. (NetWare 2.x, 3.x, 4.x)
Set to keep a file or directory from being displayed in a directory listing. (NetWare 2.x,
3.x, 4.x)
Set to make it faster to access a file with many clusters on a hard disk. (NetWare 2.x,
3.x, 4.x)
Set to make sure that a file is compressed immediately. (NetWare 2.x, 3.x, 4.x)
Automatically set to show that a file has been migrated to secondary storage medium.
(NetWare 4.x)
Set to make sure a file or directory is purged (zeroed) immediately after deletion, so
that no data from the file is available. (NetWare 3.x, 4.x)
Set to make sure a file or directory name is not changed. (NetWare 3.x, 4.x)
Supported but not used.
Set to specify whether a file can be modified. (NetWare 2.x, 3.x, 4.x)
Set to indicate that multiple users or processes can access a file simultaneously.
(NetWare 2.x, 3.x, 4.x)


Prev Page 89 Next

AUP (Acceptable Use Policy)
77
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
SEE ALSO
Access Rights
MAU (Access Unit)
In the 1988 version of the CCITT's X.400
Message Handling System (MHS), an AU
is an application process that provides a
CCITT-supported service, such as faxing,
with access to a Message Transfer System
(MTS). The MTS can deliver a message to
users or services at any location accessible
through the MHS.
AUs supplement user agents (UAs), which
give human users access to an MTS.
BROADER CATEGOR Y
X.400
COMPARE
PDAU; UA (User Agent)
M
Audio Frequency Range
The range of frequencies that the human ear
can hear, which goes from a frequency of 20 M
hertz to about 20 kilohertz (although few
people can hear the extremes well). People
can produce sounds within only a small por-
tion of this range, from about 100 to 3,000
hertz, which is the bandwidth of the ordi-
nary, acoustically-based telephone system.
M
Audit
An examination of network activity to
make sure that the network monitoring
and data gathering are working correctly.
Although this is a management activity, it
is done independently of the network man-
agement package in some environments (for
example, in NetWare). An independent
audit can check the reliability of the man-
agement software.
M
AUI (Attachment Unit Interface)
One component of the physical layer, as
defined in the IEEE 802.x specifications and
in the OSI Reference Model. The other two
components are the physical layer signaling
(PLS) above the AUI and the physical
medium attachment (PMA) below it.
SEE ALSO
Connector, AUI (Attachment Unit
Interface
AUP (Acceptable Use Policy)
An AUP represents guidelines established for
the use of the Internet or of the services from
a particular provider. For example, in the
early days, commercial traffic was not
allowed on the Internet, according to the
ATTRIBUTE
DESCRIPTION
Sy (System)
T (Transactional)
Wa (Write audit)
Set to indicate that a file or directory is a NetWare or DOS system file or directory.
(NetWare 2.x, 3.x, 4.x)
Set to allow NetWare's Transactional Tracking System (TTS) to protect a file. (Net-
Ware 2.x, 3.x, 4.x)
Supported but no AU (Access Unit)


Prev Page 90 Next

78
Authentication
NSF's (National Science Foundation) AUP.
Internet service providers may also stipulate
AUPs. For example, providers may restrict
or prohibit distribution of newsletters or
other postings to large subscriber lists.
MAuthentication
In network security and other operations,
authentication is the process of determining
the identity and legitimacy of a user, node,
or process. Various authentication strategies
have been developed. Among the simplest
are the use of user IDs and passwords.
A relatively new authentication scheme,
called digital signatures, is very effective and
almost impossible to fool (unless one has
access to the private encryption key of one
party). In digital signatures, a user (user A)
uses another user's (user B's) public key to
encrypt the transmission, and uses A's pri-
vate key to "sign" it. At the receiving end,
user B uses A's public key to validate the sig-
nature, and user B's private key to decrypt
the transmission.
The CCITT distinguishes two levels of
authentication for directory access in its
X.509 recommendations:
I Simple authentication, which uses just
a password and works only for limited M
directory domains.
I Strong authentication, which uses
a public key encryption method
to ensure the security of a
communication.
BROADER CATEGOR Y
Network Security
M
Authentication System
An authentication system is a server whose
job is to check the validity of all identities on
the network and of their requests. Most of
the work is done automatically, without
requiring any explicit human intervention.
One example of an authentication system
is Kerberos, which was created for Project
Athena at MIT. Kerberos is a distributed
authentication system which verifies that a
user is legitimate when the user logs in and
every time the user requests a service. Ker-
beros uses special keys, called tickets, to
encrypt transmissions between Kerberos
and a user.
BROADER CATEGOR Y
Network Security
MAuthority and Format Identifier (AFI)
SEE
AFI (Authority and Format Identifier)
MAutocall Unit (ACU)
SEE
ACU (Autocall Unit)
AUTOEXEC.BAT
Under DOS, AUTOEXEC.BAT is a special
batch file that is executed automatically
when the computer boots or reboots. The
commands in the file can be used to config-
ure a working environment. For example,
commands in an AUTOEXEC.BAT file may
load drivers or other files, set a command
line prompt, set environment variables, load
a network operating system, and so on.


Prev Page 91 Next

Auto-Partition Algorithm
79
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
Various solutions have been developed
to allow some flexibility in booting to an
environment. For example, OS/2 version
2.x allows each DOS process to have its
own automatically executed file. For DOS,
various programs have been developed
to allow conditional processing in the
AUTOEXEC.BAT file.
BROADER CATEGOR Y
Boot
SEE ALSO
AUTOEXEC.NCF; CONFIG.SYS
M
AUTOEXEC.NCF
On a NetWare server, AUTOEXEC.NCF is
an executable batch file that is used to con-
figure the NetWare operating system and to
load the required modules. The following
are some of the tasks of AUTOEXEC.NCF:
I Store the server name and IPX internal
network number.
I Load local-area network (LAN) driv-
ers and the settings for the network
interface cards (NICs).
I Bind protocols to the installed drivers.
I Load NetWare Loadable Modules
(NLMs).
I Set time-zone information on the
network.
I Execute certain server commands.
COMPARE
AUTOEXEC.BAT
MAutomatic Alternate Routing (AAR)
SEE
AAR (Automatic Alternate Routing)
MAutomatic Call Distributor
A device that automatically switches an
incoming call to the next available line.
MAutomatic Number Identification
(ANI)
SEE
ANI (Automatic Number Identification)
M
Automatic Repeat Request (ARQ)
SEE
ARQ (Automatic Repeat Request)
M
Automatic Rollback
In NetWare's Transaction Tracking System
(TTS), a feature that restores the starting
state of a database if a transaction fails
before completion.
MAutomatic Route Selection (ARS)
SEE
ARS (Automatic Route Selection)
MAuto-Partition Algorithm
An algorithm by which a repeater can auto-
matically disconnect a segment from a net-
work if that segment is not functioning
properly. This can happen, for example,
when a broken or unterminated cable causes


Prev Page 92 Next

80
A/UX
too many collisions. When the collisions
have subsided, the network segment can be
reconnected.
MA/UX
An implementation of the UNIX operating
system on a Macintosh, enhanced with some
Macintosh-specific features, such as support
for the Macintosh Toolbox. A/UX is based
on System V Release 2 (SVR2) of AT&T's
UNIX.
M
AUX
In DOS, AUX is the logical name for an
auxiliary device. This is usually the serial
communications board, which is more
commonly known as COM1.
M
Availability
In network performance management, the
proportion of time during which a particular
device, program, or circuit is ready for use.
Specifically, the availability of a device is the
ratio of MTBF to (MTBF + MTTR), where
MTBF and MTTR are mean time before
failure and mean time to repair, respectively.
A device is considered available even if it is
in use.
MAvalanche Photodiode (APD)
SEE
APD (Avalanche Photodiode)
MAWG (American Wire Gauge)
AWG (American Wire Gauge) is a classifica-
tion system for copper wire. The system is
based on the gauge, or diameter, of the con-
ducting wire. The lower the gauge, the
thicker the wire and the lower the resistance
per unit length. The table "Diameter and
Resistance Values for Selected Wire Gauges"
shows some gauge values and corresponding
diameters.
DIAMETER AND RESISTANCE
VALUES FOR SELECTED WIRE
GAUGES
AWG
VALUE
(GAUGE) (MM)
DIAMETER
RESISTANCE
(OHMS/
METER)
30
24
22
20
18
16
14
12
0.26
0.51
0.64
0.81
1.02
1.29
1.63
2.05
0.346
0.080
0.050
0.032
0.020
0.012
0.008
0.005


Prev Page 93 Next

BB


Prev Page 94 Next

82
B8ZS (Bipolar with 8 Zero Substitution)
BM
B8ZS (Bipolar with 8 Zero
Substitution)
A signal-encoding scheme in which a 1 is
represented alternately as positive and nega-
tive voltage, and 0 is represented as zero
voltage. B8ZS requires at least one bit of
every eight to be a 1.
SEE ALSO
Encoding, Signal
MBAC (Basic Access Control)
In the CCITT X.500 directory services
model, the more comprehensive of two sets
of access-control guidelines. The less com-
prehensive set is called SAC (Simplified
Access Control).
SEE ALSO
X.500
MBackbone
In a hierarchically arranged distributed sys-
tem, the backbone is the top-level, or cen-
tral, connection path shared by the nodes or MBack End
networks connected to it.
The backbone manages the bulk of the
traffic, and it may connect several different
locations, buildings, and even smaller net-
works. The backbone often uses a higher-
speed protocol than the individual local-area
network (LAN) segments.
M
Backbone Network
A backbone network is one with a central
cabling scheme (the backbone) to which
other networks are attached. Nodes in one
network can talk to nodes in other networks
by sending packets across the backbone
network.
The networks attaching to the backbone
are known as access networks. Access net-
works may require a gateway or router to
attach to the backbone network.
A backbone network can be useful in
decentralized corporations. For example, a
backbone network might be used in a com-
pany in which each department has set up its
own network and several different architec-
tures are used. Since the backbone network
leaves the access networks intact, those net-
works can continue operating as if they were
not on the larger network. However, the
backbone gives each of the networks access
to the resources and data of the other access
networks.
One obstacle to a successful backbone
network is the high bandwidth that may be
required to handle potentially heavy traffic.
Because of this consideration, fiber-optic
cable is the most sensible cabling for back-
bone networks.
In a client/server architecture, the portion of
an application that runs on the server and
does the actual work for the application.
The front end runs on the client machine
and provides an interface through which the
user can send commands to the back end.
M
Background Process
A process or program that executes inciden-
tally, while another process or program is
operating in the foreground. The foreground


Prev Page 95 Next

Backup
83
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
process gets the main attention of the CPU
(central processing unit), and the back-
ground process takes CPU cycles when the
foreground process is temporarily idle.
MBacking Out
In NetWare's TTS (Transaction Tracking
System), the process of abandoning an
uncompleted database transaction, leaving
the database unchanged. TTS takes this
action to ensure that the database is not cor-
rupted by information from an incomplete
transaction.
SEE ALSO
TTS (Transaction Tracking System)
M
Backplane
A backplane is a circuit board with slots into
which other boards can be plugged, as illus-
trated in the figure "A backplane." The
motherboard in a PC is a backplane.
A segmented backplane is a backplane
with two or more buses, each with its own
slots for additional boards.
M
Backplate
The metal bracket at one end of a circuit
board, usually at the back when the board is
plugged into an expansion slot. The back-
plate, also known as an end bracket or
mounting bracket, typically has cutouts for
connectors and switches. PCs usually come
with blank backplates over each expansion
slot, which are removed when you plug a
board into the slot.
MBackscattering
In a fiber-optic transmission, light that is
reflected back in the direction from which
the light came.
M
Backup
A backup is an archival copy that is stored
on an external medium. For example, a
backup might contain the contents of a hard
disk or a directory.
The creation of regular backups is essen-
tial in a networking environment. An effec-
tive backup system ensures that data stored
on the network can be recreated in the event
of a crash or another system failure.
Networking packages differ in the type of
backup supported, in the media to which
material can be backed up, and in the ease
with which parts of the archived material
can be restored. Backups are generally made
to tape or to erasable optical (EO) media.
No serious network should be backed up
to floppy disks.
Various types of backups are distin-
guished, including full, differential, and
incremental. In full backups, a copy is made
of all the data.
In differential and incremental backups,
only the data that has been added or
changed since the previous backup is
included. Differential and incremental back-
ups assume a full backup has been done and
they merely add to this material. Such back-
ups use the Archive flag (attribute), which is
supported by DOS and most networking
environments. This flag is associated with a
file and is set whenever the file is changed
after the file is backed up.


Prev Page 96 Next

84
Backup
The backed up material should generally
be stored in a different physical location
from the original material, and should be
protected from disasters such as fire, flood,
magnets, theft, and so on.
Backup operations should be done at a
time when the network is not being used for
its ordinary activity, which generally means
outside regular working hours. One reason
for this is that most backup programs will
not back up a file that is open. Truly, the
work of a system administrator is never
done.
When you restore the data, you restore
the last full backup first, then restore each
incremental backup made since the last full
backup.
A BACKPLANE
486 DX
Jumpers
Expansion
Slots
Keyboard Connector
Slot
for Optional
Memory
Card
Memory Chip
Slots
Coprocessor
CPU
Dip Switches
Power Connectors
BIOS
Chips


Prev Page 97 Next

Balun
85
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
SEE ALSO
Archive
RELATED AR TICLES
Data Protection; Disk Duplexing; Disk
Mirroring

M
Backward Error Correction (BEC)
SEE
BEC (Backward Error Correction)
MBad-Block Revectoring
In data protection, the process by which
material written to a defective area of the
hard disk is retrieved and rewritten to a dif-
ferent, nondefective area of storage. The
defective area is identified as such in a bad
block table, so that future writes will not be
made to the area. Bad-block revectoring is
known as a Hot Fix in Novell's NetWare.
BACKUP TIPS
I Keep multiple copies of backups; redundancy
should be a part of your backup plan.
I Test your backups to make sure that they are
what you think they are.
I Store your backups in a secure, off-site
location.
I Replace your backup media on a regular basis.
I Consider making incremental backups of
critical data at more frequent intervals.
M
Bad Block Table
In storage management, a table in which all
known defective areas of a hard disk are
listed to ensure that nothing will be written
to these areas. The process of protecting
data in this manner is known as bad-block
revectoring, or Hot Fix in Novell's NetWare.
M
Balun
A balun is a hardware device used to
adjust impedances in order to connect differ-
ent types of cable. The name comes from
balanced/unbalanced, because the device is
often used to connect twisted pair (bal-
anced) to coaxial (unbalanced) cable.
Baluns may have different connectors at
each end to make them compatible with the
cable types being connected. For example, a
balun might have a BNC connector at one
end and an RJ-45 connector at the other.
A balun makes it possible to use twisted-
pair wiring that may already be installed in
parts of a building or office in conjunction
with coaxial cable that is coming from else-
where or that has been installed more
recently. The balun controls the electrical
signal's passage from one cable type to the
other, but does not change the signal in any
other way. Similarly, a balun enables you to
connect a network interface card designed
for use with coaxial cables to a hub that uses
twisted-pair cabling.
Baluns vary with respect to the cable
gauge (thickness) supported and to the max-
imum cable distance over which the signal is
supported. This distance may be as high as
360 to 460 meters (1,200 to 1,500 feet).
Coaxial boosters may be used to increase


Prev Page 98 Next

86
Bandwidth
signal strength in the coaxial cable, and thus
increase the distance over which the signal
will be supported by the balun. However,
such boosters can cost up to ten times as
mach as a balun, and will only double the
supported distance.
BROADER CATEGORIES
Connector; Intranetwork Link
M
Bandwidth
Bandwidth refers to the amount of data a
cable can carry; measured in bits per second
(bps) for digital signals, or in hertz (Hz) for
analog signals such as sound waves. An ana-
log bandwidth is computed by subtracting
the lower frequency from the higher one.
For example, the bandwidth of the human
voice is roughly 2,700 Hz (3,000 - 300).
A larger bandwidth means greater poten-
tial data-transmission capability. For digital
signals, a higher bit rate represents a larger
bandwidth. However, the higher the fre-
quency, the shorter the wavelength. A higher
bandwidth (that is, a higher signal fre-
quency) means faster transmission, which
means a shorter signal. With a short signal,
there is a smaller margin for error in inter-
preting the signal. This means that the
effects of attenuation and other signal dis-
tortion must be kept to a minimum.
A signal traveling along a cable degrades
with distance. It is possible to connect the
cable to special components that can clean
up and rejuvenate a signal. High-frequency
electrical signals must be cleaned up
WHAT TO LOOK FOR IN A BALUN
Baluns may include a stretch of cable (at extra cost, of course). Here are some things to consider when you're
shoppping for a balun:
I Baluns work most reliably when the cable has low capacitance (20 picofarads/foot or less) and when the
cable impedance is not too high.
I Baluns are available in different qualities, based on the type and gauge (thickness) of cable at either end.
Make sure the balun you select supports the cable properties and distances you need and then some. To be
on the safe side, don't use a balun (or any other kind of connector, for that matter) at the maximum rated
length.
I
Some network interface card manufacturers recommend specific baluns for their boards. Similarly, some
manufacturers suggest that you do not use baluns with their hubs or cards. Check with the manufacturer to
determine whether either is the case with the network interface card or hub you plan to use.
I When using a balun on a network, you'll almost certainly want a balun designed for data transmission,
because this type is made for direct (rather than reversed) pin-to-pin connections.
I Baluns pass signals on, so the balun's reliability depends on the signal's quality. For this reason, it's not a good
idea to use a balun with passive hubs, which don't clean and strengthen the signal before passing it on.


Prev Page 99 Next

Bang Path
87
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
frequently, which means single cable seg-
ments must be short.
Some commonly used frequency bands
for analog transmissions are shown in the
table "Bandwidths on the Electromagnetic
Spectrum."
Very low frequency (VLF) through super
high frequency (SHF) are considered the
radio spectrum. The bandwidths are used as
follows:
I AM radio broadcasts in the medium
frequency (MF) range (535 to 1,605
kHz).
I FM radio and VHF television broad-
cast in the very high frequency (VHF)
range (88 to 108 MHz for FM; the
split ranges from 54 to 88 MHz and
from 174 to 216 MHz for VHF
television).
I Cable stations broadcast over several
bands (frequency ranges) in the VHF
and ultra high frequency (UHF) ranges M
(108 to 174 MHz in the VHF range;
216 to 470 MHz in the VHF and UHF
ranges).
I UHF television broadcasts in the UHF
range (470 to 890 MHz).
I
Radar operates at 10 different bands
over a huge frequency range (230
MHz to 3 THz).
Radio Spectrum Bandwidths
For digital transmissions, bandwidths range
considerably. Here are some examples of
bandwidth values for digital transmissions:
I Some digital telephone lines: less than
100 kbps
I ARCnet networks: 2.5 Mbps
I
ARCnet Plus networks: 20 Mbps
I Ethernet networks: 10 Mbps
I Fast Ethernet networks: 100 Mbps
I Token Ring networks: 1, 4, or 16
Mbps
I Fast Token Ring networks: 100 Mbps
I Fiber-optic (FDDI) networks: About
100 Mbps, but can theoretically be
several orders of magnitude higher
I ATM networks: about 655 Mbps, with
speeds as high as 2.488 gigabits per
second (Gbps) in the future
Bang Path
On the Internet, a bang path is a series of
names that specifies a path between two
nodes. A bang path is used in uucp (UNIX-
to-UNIX copy program) and sometimes for
e-mail (electronic mail) or communications
on BITNET. The path consists of domain or
machine names separated by exclamation
points (!), known as bangs in some comput-
ing circles. For example, in a bang path such
as hither!thither!yon, hither might be a gate-
way, thither a computer, and yon a user.
Bang paths go back to the days before
automatic routing, because explicit paths
Digital Transmission Bandwidths


Prev Page 100 Next

88
Bang Path
BANDWIDTHS ON THE ELECTROMAGNETIC SPECTRUM
NAME
BANDWIDTH
(FREQUENCY
RANGE)
WAVELENGTH
COMMENTS
Ultra-low frequency (ULF)
Extra low frequency (ELF)
Voice frequency (VF)
Very low frequency (VLF)
Low frequency (LF)
Medium frequency (MF)
High frequency (HF)
Very high frequency (VHF)
Ultra-high frequency (UHF)
Super high frequency (SHF)
Extremely high frequency (EHF)
Infrared(IR)
Visible
Ultraviolet (UV)
X-ray
.001 Hz (hertz)1 Hz
30 Hz300 Hz
300 Hz3 kHz (kilohertz)
3 kHz30 kHz
20 kHz100 kHz
30 kHz300 kHz
300 kHz3 MHz
3 MHz30 MHz
30 MHz300 MHz
300 MHz3 GHz
3 GHz 30 GHz
30 GHz300 GHz
300GHz300THz
300 GHz430 THz
430 THz 750 THz
750 THz30 PHz
(petahertz, or quadril-
lions of hertz; a quadril-
lion is 1015, or roughly
250)
30 PHz30 EHz
(exahertz, or quintillions
of hertz; a quintillion is
1018, or roughly 260)
300 Gm (gigameter, or
billions of meters)-
300 Mm (megameter, or
millions of meters)
10 Mm1 Mm
1 Mm100 km
(kilometer)
100 km10 km
150 km30 km
10 km1 km
1 km100 m
100 m10 m
10 m1 m
1 m10 cm
10 cm1 cm
1 cm1 mm
1 mm1 micron
1 mm0.7 micron
0.7 micron0.4 micron
400 nm - 10 nm
10 nm0.01 nm
Subsonic
Audible spectrum
Ultrasonic
Long wave
Medium wave
Ultra-shortwave
Ultramicrowave
Visible spectrum
Ultraviolet
X-ray


Prev Page 101 Next

Basic Access Control (BAC)
89
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
were needed when sending to or communi-
cating with another location.
M
Banner Page
A banner page is output by a printer in a
network environment to separate print jobs.
A banner page is also known as a job sepa-
rator page. Printing of this page is controlled
by the network operating system.
A banner page might indicate the name of
the user who printed the file and other infor-
mation. You can eliminate banner pages in
NetWare and in most other network operat-
ing systems.
M
Base Address
In memory allocation, a base address defines
the starting or reference location for a block
of contiguous memory. The memory may be
general-purpose, or it may serve as cache or
port memory. Here are some examples of
different types of base addresses:
I A base I/O (input/output) address is
the starting location for the memory
area allocated for an I/O port. The
processor uses this address to find the
correct port when the processor needs
to communicate with a device.
I A base memory address is the starting
location for a block of memory, such
as a buffer area.
I A base video address is the starting
location for video memory.
MBaseband
In networking, a baseband connection is one
that uses digital signals, which are sent over
wires without modulation; that is, binary
values are sent directly as pulses of different
voltage levels rather than being superim-
posed on a carrier signal (as happens with
modulated transmissions). Baseband net-
works can be created using twisted-pair,
coaxial, or fiber-optic cable.
Even though only a single digital stream
is transmitted over a baseband connection,
it is possible to transmit multiple signals.
This is done by multiplexing (combining
several signals in a transmission by inter-
leaving the signals using, for example,
time slices).
This digital signaling is in contrast to
broadband, in which analog signals are sent
over multiple channels at the same time.
Each channel is allocated a different fre-
quency range.
M
Baseline
In performance analysis, a reference level or
the process of determining this level. For
example, in a networking context, a baseline
measures performance under what is consid-
ered a normal load. Commonly used base-
line measures include transmission rate,
utilization level, and number of lost or erro-
neous packets.
M
Basic Access Control (BAC)
SEE
BAC (Basic Access Control)


Prev Page 102 Next

90
Basic Information Unit (BIU)
M
Basic Information Unit (BIU)
SEE
BIU (Basic Information Unit)
M
Basic Link Unit (BLU)
SEE
BLU (Basic Link Unit)
MBasic Mode
In an FDDI II network, a mode of operation
in which data can be transmitted using
packet-switching. This is in contrast to
hybrid mode, in which both data and voice
can be transmitted.
SEE ALSO
FDDI (Fiber Distributed Data Interface)
M
Basic Rate Access (BRA)
SEE
BRA (Basic Rate Access)
MBasic Telecommunications Access
Method (BTAM)
SEE
BTAM (Basic Telecommunications Access
Method)
MBasic Transmission Unit (BTU)
SEE
BTU (Basic Transmission Unit)
M
Baud Rate
The baud rate is the measure of the number
of times an electrical signal can be switched
from one state to another within a second.
The faster a switch can occur, the higher the
baud rate.
The relationship between baud and bit
transfer rates depends on the number of bit
values that are encoded in a single signal.
When each signal represents one bit, the bit
and baud rates are equal; when a signal
encodes multiple bits, the bit rate is a multi-
ple of the baud rate.
The term baud comes from Baudot, the
name of a French telegraph operator who
developed a five-bit encoding system in the
late 19th century. This Baudot code is still
used, officially known as International Tele-
graph Alphabet #1.
Since it is a violation of the bylaws for
workers in computers and communications
to pass up an opportunity to create an acro-
nym, the term also doubles as the acronym
for bits at unit density.
COMPARE
Bit Rate
M
BBS (Bulletin Board System)
A BBS is one or more computers set up with
modems so that users can access those com-
puters from remote locations. Users dialing
into the BBS can send messages, get techni-
cal support from a vendor, upload or down-
load files, and so on.
Many BBSs are set up by vendors to
provide users with a forum for communica-
tion and with delayed access to technical


Prev Page 103 Next

Bellman-Ford Algorithm
91
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
support. Some BBSs are set up to provide
services to a specialized market, generally
for a fee. (Fee-based BBSs are often given
more aggrandized names, such as Informa-
tion Services.)
M
BCC (Block Check Character)
In longitudinal redundancy checks (LRCs), a
character inserted at the end of a block to
provide error-detection capabilities. Each of
the character's bits is a parity bit for a col-
umn of bits in the block.
SEE ALSO
CRC (Cyclic Redundancy Check)
MBCD (Binary Coded Decimal)
An encoding scheme in which each digit is
encoded as a four-bit sequence.
M
B Channel
In an ISDN system, the bearer channel that
carries voice or data at 64 kilobits per sec-
ond in either direction. This is in contrast to
the D channel, which is used for control sig-
nals and data about the call. Several B chan-
nels can be multiplexed into higher-rate H
channels.
SEE ALSO
BRI (Basic Rate Interface); PRI (Primary
Rate Interface)
MBCN (Beacon)
A frame used in a token ring network to
indicate that a hard error (one that is serious
enough to threaten the network's continued
operation) has occurred in the node sending
the beacon frame or in this node's nearest
addressable upstream neighbor (NAUN).
M
BCP (Byte-Control Protocols)
Protocols that are character- (rather than
bit) oriented.
MBEC (Backward Error Correction)
Error correction in which the recipient
detects an error and requests a retransmis-
sion. The amount of material that needs to
be retransmitted depends on the type of con-
nection, how quickly the error was detected,
and the protocols being used.
COMPARE
FEC (Forward Error Correction)
M
Bel
A bel is a unit for measuring the relative
intensity of two levels for an acoustic, elec-
trical, or optical signal. The bel value is
actually proportional to the logarithm (to
base 10) of this ratio.
For example, if one voltage is 10 times as
strong as another, the higher voltage is one
bel higher than the lower one; similarly, if
one sound is 100 times as loud as another,
the louder sound is two bels louder. The
decibel, a tenth of a bel, is used more com-
monly when computing such values.
M
Bellman-Ford Algorithm
An algorithm for finding routes through an
internetwork. The algorithm uses distance
vectors, as opposed to link states. The


Prev Page 104 Next

92
BER (Basic Encoding Rules)
Bellman-Ford algorithm is also known
as the old ARPAnet algorithm.
SEE ALSO
Algorithm
M
BER (Basic Encoding Rules)
In the ISO's Abstract Syntax Notation One
(ASN.1), the BER are the rules for encoding
data elements. Using the BER, it is possible
to specify any ASN.1 element as a byte
string. This string includes three compo-
nents, and the encoding may take any of
three forms, depending on the information
being encoded.
The components of BER are the Type,
Length, and Value fields.
The Type, or identifier, field indicates the
class of object, as well as the string's form.
Examples of ASN.1 types include BOOL-
EAN, INTEGER, BIT STRING, OCTET
STRING, CHOICE, and SEQUENCE OF.
Of these, the first two are primitive, the next
three may be primitive or constructed types,
and the SEQUENCE OF type is always con-
structed. (A primitive object consists of a
single element of a particular type of
information, such as a number or logical
value; a constructed type is made up of other
simpler elements, such as primitive objects
or other constructed types.)
The Length field indicates the number of
bytes used to encode the value. Values actu-
ally may have a definite or an indefinite
length. For the latter case, a special value is
included in the last byte.
The Value, or contents, field represents
the information associated with the ASN.1
object as a byte string. For primitive types,
this is a single value; for constructed types,
there may be several values, possibly of dif-
ferent types, involved.
M
BER Encoding
The encoding may be any of the following:
I Primitive/fixed length, which consists
only of a primitive object and which is
always a fixed length. For example, an
integer variable is of this type.
I Constructed/fixed length, which con-
sists of a group of objects and values,
with a fixed total length. For example,
this might be a record with only pre-
defined components, all of which have
a fixed and known length.
I Constructed/variable length, which
consists of a group of objects whose
total size may vary from case to case,
so that a special value is needed to
indicate the end of the value.
The BER can provide an encoding for any
valid ASN.1 object. One difficulty is that the
rules can sometimes provide more than one.
In this case, the rules may be too general,
because all the "synonymous" rules eat up
overhead.
M
BER Variants
Several variants of the BER have been
proposed and are being developed. In gen-
eral, these are designed to provide faster,
simpler, and/or more generic encodings. The
Components of BER


Prev Page 105 Next

Big-Endian
93
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
following are some of the alternatives that
have been proposed:
I CER (canonical encoding rules), which
represent a subset of the BER. With the
canonical rules, it should be possible
to eliminate any redundant paths,
which can slow down performance
considerably.
I DER (distinguished encoding rules),
which are also a subset of BER.
I LWER (lightweight encoding rules),
which make faster encoding possible,
but may result in larger transmissions.
I PER (packed encoding rules), which
are used to compress the information
about an object.
PRIMAR Y SOURCES
CCITT recommendation X.209; ISO
document 8825
BROADER CATEGOR Y
ASN.1
M
BER (Bit Error Rate)
Number of erroneous bits per million (or
billion or trillion) bits in a transmission or a
transfer (as from a CD to memory). The
BER depends on the type and length of
transmission or on the media involved in a
transfer.
COMPARE
BLER (Block Error Rate)
MBerkeley Internet Name Domain
(BIND)
SEE
BIND (Berkeley Internet Name Domain)
M
BERT (Bit Error Rate Tester)
A hardware device for checking a transmis-
sion's bit error rate (BER), or the proportion
of erroneous bits. The BERT sends a pre-
defined signal and compares it with the
received signal. BERTs are moderately
expensive devices that are used most com-
monly for troubleshooting wiring.
COMPARE
BLERT (Block Error Rate Tester)
M
BIA (Burned-In Address)
A hardware address for a network interface
card. Such an address is assigned by the
manufacturer and is unique for each card.
M
BIB (Bus Interface Board)
An expansion board. In particular, a net-
work interface card (NIC), which serves as
an interface between the node (computer)
and the network medium.
MBig-Endian
In data transmission and storage, the order
in which bytes in a word are processed
(stored or transmitted). The term comes
from Jonathan Swift's Gulliver's Travels,
in which a war is fought over which end of
an egg should be cracked for eating. This


Prev Page 106 Next

94
BIND (Berkeley Internet Name Domain)
ordering property is also known as the pro-
cessor's byte-sex.
In big-endian implementations, the high-
order byte is stored at the lower address.
Processors in mainframes (such as the IBM
370 family), some minicomputers (such as
the PDP-10), many RISC machines, and also
the 68000 family of processors use big-
endian representations. The IEEE 802.5
(token ring) and the ANSI X3T9.5 FDDI
standards use big-endian representations. In
contrast, the 802.3 (Ethernet) and 802.4
(token bus) standards use little-endian
ordering.
The term is used less commonly to refer
to the order in which bits are stored in a
byte.
COMPARE
Little-Endian; Middle-Endian
M
BIND (Berkeley Internet Name
Domain)
In the Internet community, a domain name
system (DNS) server developed at the Uni-
versity of California, Berkeley, and used on
many Internet machines.
M
Bindery
In Novell's NetWare products, the bindery is
a database maintained by the network oper-
ating system (NOS) on each server. The
bindery is located in the SYS:SYSTEM direc-
tory and contains information about all the
users, workstations, servers, and other
objects recognized by the server.
The bindery information determines the
activities possible for the user or node. In
the bindery, this information is represented
as a flat database.
The bindery has three types of
components:
Objects: Users, devices, workgroups,
print queues, print servers, and so on.
Most physical and logical entities are
regarded as objects.
Properties: Attributes, specifically, as
assigned to bindery objects, such as
full name, login restrictions, or group
membership information.
Property data sets: The values that will
be stored in an object's property list.
The bindery has been replaced in Net-
Ware 4.x by the NetWare Directory Services
(NDS), in which information is represented
hierarchically in tree format.
However, version 4.x includes bindery-
emulation capabilities, which makes it possi-
ble to integrate bindery-based objects into a
network based on NDS. In NetWare 4.1, the
Bindery services utility creates a bindery
context within which the bindery objects
appear as a flat database-as required by
earlier versions of NetWare. This perspective
is valid in only a limited context, which
makes it possible to integrate the bindery
information into the NDS while still provid-
ing a pre-4.x server with access to the bind-
ery's contents.
Another 4.1 utility, NetSync, makes it
possible to manage up to 12 NetWare 3.x
servers within a NetWare 4.1 network. This
makes all 12 servers look like a single server
to users-a user would need only one login
to access as many of the NetWare 3.x servers


Prev Page 107 Next

BIOS Extensions
95
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
as desired. (As always, such access assumes
that the user has the necessary privileges.)
With NetSync, it also becomes easier to
update resources on different machines.
BROADER CATEGOR Y
NetWare
M
Bindery Emulation
In Novell NetWare 4.x, bindery emulation is
a NetWare Directory Service that makes the
Directory database emulate a flat database.
In NetWare 2.x and 3.x, information
about all network objects is stored in a
flat database, called the bindery. A flat data-
base is one in which all objects in the
database exist as entities of equal standing;
an object cannot contain another object.
In NetWare 4.x, network objects and their
related information are contained in a hier-
archical database called the Directory. A
hierarchical database can contain several
levels of objects, which means that objects
can contain other objects.
Bindery emulation allows programs that
were written to run under the NetWare
bindery to find the network object informa-
tion they need in NetWare 4.x's Directory by
making the information in the Directory
appear as a flat structure.
Such bindery emulation is provided
by the Bindery services utility, which makes
the bindery's contents look appropriate
for whatever server is querying it (i.e, 3.x
or 4.x).
BROADER CATEGOR Y
NetWare
MBinding and Unbinding
In a local-area network (LAN), binding is
the process of associating a communication
protocol, such as TCP/IP, IPX/SPX, or
AppleTalk, and a network interface card
(NIC). Unbinding is the process of dissociat-
ing the protocol from the NIC.
The LAN driver for a card must have at
least one communication protocol associ-
ated with it. The LAN driver will be able to
process only those packets that use the asso-
ciated protocol.
M
BIOS (Basic Input/Output System)
The BIOS is a collection of services on a
ROM (read-only memory) chip. The BIOS
services enable hardware and software,
operating systems and applications, and also
applications and users to communicate with
each other. The BIOS services are loaded
automatically into specific addresses and
should always be accessible.
BIOS services are updated and expanded
to handle newer devices and greater
demands. To get a newer BIOS, you simply
need to replace the ROM chip in your com-
puter with an appropriate upgrade chip.
M
BIOS Extensions
A collection of services that supplement
those provided by the standard BIOS (Basic
Input/Output System). Like the standard
BIOS, BIOS extensions are implemented on
a ROM (read-only memory) chip, located
on the motherboard or on an expansion
board.


Prev Page 108 Next

96
Bipolar with 8 Zero Substitution (B8ZS)
M
Bipolar with 8 Zero Substitution
(B8ZS)
SEE
B8ZS (Bipolar with 8 Zero Substitution)
BISDN is an extension of the ISDN (Inte-
grated Services Digital Network) to allow
multiple types of information to be transmit-
ted. BISDN can handle voice, video, and
graphics, as well as data.
Whereas ISDN networks generally use
some form of time division multiplexing
(TDM) for actual transmissions, BISDN net-
works generally use ATM (asynchronous
transfer mode) as their transmission technol-
ogy. ATM is often erroneously regarded as
being equivalent to BISDN.
Figure "BISDN Services" summarizes the
kinds of capabilities that have been defined
for BISDN networks. The services are
grouped into two main groups, each with
multiple service classes:
I Interactive services are those in which
the user can initiate the service and
influence its direction. Three classes
are distinguished, and each class
includes several examples. For exam-
ple, conversational services include
video-conferencing and video-
telephony (for shopping, learning,
etc). Online research is included
among interactive services.
BISDN (Broadband ISDN)
BISDN Services
I Distribution services are those in
which information (in the form of
video, documents, or data) can be
broadcast to whomever has the
resources and rights to receive the
broadcast. Distribution services are
divided into those for which the user
has no control over the presentation
(other than to turn it on or off) and
those where the user can control which
elements are received. Examples of the
former include TV programming and
electronic newspapers; examples of the
latter include retrieval of selected news
items and certain online courses.
PRIMAR Y SOURCES
BISDN is discussed in more than a few of
the documents in the ITU-T I.xxx docu-
ment series. For example, I.113 provides
a vocabulary for BISDN, and I.121 pro-
vides a list of the documents that discuss
BISDN or ATM or both. These include
I.150 (ATM for BISDN), I.211 (BISDN
services), I.311 (General BISDN network-
ing aspects), I.327 (BISDN functional
architecture), I.361, I.362, and I.363
(ATM layers), I.413 and I.432 (BISDN
User-network interface), and I.610 (Oper-
ation and maintenance for BISDN). In
some cases, these recommendations must
be read in relation to their ISDN counter-
parts, whose numbers are generally lower
than the corresponding BISDN docu-
ment. For example, I.210 discusses ISDN
services.
COMPARE
ISDN (Integrated Services Digital
Network)


Prev Page 109 Next

Bit Error Rate Tester (BERT)
97
a
b
c
d
e
f
g
hi
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
M
Bit
A binary digit; the smallest unit of infor-
mation. A bit can have a value of 0 or 1
in a digital system. All but the low-level
protocols move information in larger
chunks, such as bytes, which consists of
multiple bits.
M
Bit Error Rate (BER)
SEE
BER (Bit Error Rate)
M
Bit Error Rate Tester (BERT)
SEE
BERT (Bit Error Rate Tester)
BISDN SERVICES
..without individual user
presentation control

Video (e.g., cable and
extended cable TV, high-
definition TV, pay-TV)

Text, graphics, and ima-
ges (e.g., document
distribution)

..with individual user
presentation control

Text, graphics,
sound, and images
(e.g., remote education,
tele-advertising,
telesoftware)

Interactive Services
Distribution Services
Conversational Service
Video (e.g., videoconferencing, video
surveillance)

Sound

Data (information, files, teleaction
telemetry, alarms, etc.)

Documents (high-speed fax, images)
Messaging Service
Video mail
Document mail
Retrieval Service
Videotex
Video, document, and data retrieval


Prev Page 110 Next

98
Bit Interval
M
Bit Interval
Bit interval, also known as bit time, refers to
the amount of time a digital signal is left at a
particular voltage level to indicate a value.
Usually, the level will indicate the value of a
single bit, but it is possible to encode more
than a single bit in a voltage level, thereby
transmitting more than one bit in a single bit M
interval.
In general, the longer the bit interval, the
slower the transmission rate. For example,
when encoding a single bit at a time, a bit
interval of .01 second means a transmission
rate of only 100 bits per second (bps).
RELATED AR TICLES
Bit Rate; Encoding, Signal
MBITNET (Because It's Time Network)
BITNET is a computer network that con-
nects many educational institutions in North
America and Europe. BITNET was set up
through EDUCOM, a nonprofit educational
consortium. It is designed to provide com-
munication facilities and easy access to
files-even from remote locations-provided
that the user has the appropriate access priv-
ileges. Today, BITNET connects more than
1,000 locations.
Partly because the early nodes were pre-
dominantly IBM mainframes, BITNET still
uses the RSCS (Remote Spooling Communi-
cations Subsystem) and NJE (Network Job
Entry) protocol suites. Because of this, a
gateway is needed to communicate with
other networks, such as the Internet.
Once a gateway between the Internet and
BITNET is known, it is relatively easy to
send a message to a user on BITNET from
most Internet installations. An address such
as user@computer.bitnet will suffice, because
most Internet mail programs recognize
bitnet as a pseudo domain name.
In Canada, BITNET is known as
NetNorth, and in Europe it is known as
EARN (for European Academic Research
Network).
Bit Rate
Bit rate is a measure of throughput, or rate
of data transfer. It represents the number of
bits that are transmitted within a second in a
digital communication, measured in bits per
second (bps). The faster the bit rate, the
shorter the bit interval (the interval to signal
a bit value). For example, at a bit rate of
5,000 bps, each bit interval can be at most
.0002 second when a single bit is transmit-
ted in each bit interval.
Bit rate is often used interchangeably with
baud rate, but these two measurements are
not exactly the same. Baud rate refers to the
number of electrical signal transitions made
in a second. If a single bit is encoded in each
signal, the bit rate and baud rate will be
equal. However, if multiple bits are encoded
in a single signal, the bit rate will be higher
than the baud rate.
M
Bit Stuffing
In data transmission, a technique for ensur-
ing that specific bit patterns do not appear
as part of the data in a transmission. For
example, if six consecutive 1 values are
encountered in the transmitted data, a 0 bit
would be inserted after the fifth consecutive
1 bit. The receiver removes any inserted bits
when processing the transmission.


Prev Page 111 Next

Block
99
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
M
BIU (Basic Information Unit)
In SNA network communications, a
packet of information created when the
transmission control layer adds a request/
response header (RH) to a request/response
unit (RU). This unit is passed to the path
control layer.
SEE ALSO
SNA (Systems Network Architecture)
M
BIU (Bus Interface Unit)
An adapter card. In particular, a network
interface card (NIC), which acts as an
interface between a node (computer) and
the network.
M
BIX (BYTE Information Exchange)
BYTE Magazine's commercial online infor-
mation service. BIX provides the usual
gamut of mail, news, and entertainment ser-
vices, as well as Internet access-including
e-mail (electronic mail), ftp (file transfer
protocol), and telnet services. In addition
to a base monthly fee (which depends on the
amount of access requested), costs for Inter-
net use include access and storage charges.
FOR INFORMATION
Call 800-695-4775; 617-354-4137. You
can use telnet to access BIX over the
Internet. To do this, telnet to
x25.bix.com.
MBlackout
A total loss of electrical power. Blackouts
can be caused by cut or broken power lines,
lightning strikes, and other natural and
man-made disasters.
SEE ALSO
Power Disturbances
M
BLER (Block Error Rate)
In communications, an error rate based on
the proportion of blocks with errors. Com-
pare it with BER (bit error rate), which is
based on the number of erroneous bits
per million (or billion or trillion) bits in
a transmission.
M
BLERT (Block Error Rate Tester)
A hardware device for determining a trans-
mission's block error rate (BER), which is
the proportion of blocks with erroneous
bits. This device is also known as a BKERT.
M
Block
A block is an area of memory or storage
with a fixed size. A network operating sys-
tem block can be anywhere from 4 to 64
kilobytes (KB). DOS blocks are typically a
multiple of 2 KB. NetWare blocks are typi-
cally 4 KB. However, the actual block size
depends on the size of the volume on which
storage is being allocated.
In some environments, such as in
NetWare, a block represents the smallest
chunk of storage that can be allocated at a
time. (In NetWare, you can accept the sug-
gested block size, which is based on the size
of the volume, or you can specify the block
size you want to use.)


Prev Page 112 Next

100
Block Error Rate (BLER)
Two types of blocks are distinguished:
Disk-allocation block: Used to store
network data, at least temporarily.
Directory-entry block: Used to store
directory information.
NetWare 4.x supports block suballoca-
tion, in which a block can be broken into
512-byte chunks. These chunks can be used
to store the ends of several files. For exam-
ple, with a 4 kilobyte (KB) block size, three
5 KB files would fit into four blocks. Each of
the files would use one block and two 512-
byte chunks in the fourth block. In contrast,
these files would require six blocks (two per
file) in NetWare 3.x.
M
Block Error Rate (BLER)
SEE
BLER (Block Error Rate)
MBlock Error Rate Tester (BLERT)
SEE
BLERT (Block Error Rate Tester)
M
BLU (Basic Link Unit)
In IBM's SNA (Systems Network Architec-
ture) networks, a block, or packet, of infor-
mation at the data-link layer.
SEE ALSO
SNA (Systems Network Architecture)
MBlue Book Ethernet
Ethernet version 2.0. This term is sometimes
used to distinguish Ethernet 2.0 from the
similar, but not identical, Ethernet variant
defined in the IEEE 802.3 standard.
SEE ALSO
Ethernet
MBookmark
In gopher environments on the Internet, a
bookmark is used to mark a specific menu
or directory on a gopher server. Once the
bookmark has been created and placed at
the desired location, it's possible to get
almost immediate access to that location,
rather than having to work your way
through layers of menus.
SEE ALSO
Gopher
M
Boot
The process by which a computer is started
up and its operating system kernel is loaded
into RAM (random-access memory) is called
the boot, or bootstrap, process. Although
the details may differ when booting to dif-
ferent disk operating systems or network
operating systems, the basic steps are the
same:
I Execute a hardware self-test.
I Look in a predefined place for the boot
sector and load this code.
I Execute the boot sector program to
load other programs.
I Execute these programs to load still
other programs or to configure the
operating environment.


Prev Page 113 Next

BRI (Basic Rate Interface)
101
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
I Repeat the previous step as often as
dictated by the programs being loaded
and by their initialization code.
M
BOOTCONF.SYS
In Novell NetWare, a configuration file that
specifies how a diskless workstation can
boot the operating system in order to access
the network.
M
Boot ROM
A ROM (read-only memory) chip used in
diskless workstations to enable these
machines to boot and connect to a network.
M
Bounce
A term for the action of returning an unde-
liverable e-mail message. In such a case, the
postmaster on the system returns the mes-
sage, along with a bounce message, to the
sender.
MBRA (Basic Rate Access)
Access to an ISDN Basic Rate Interface
(BRI), an interface with two 64 kilobits per
second (kbps) B channels (for voice and
data) and one 16 kbps D channel (for call
and customer information). Compare it with
PRA, which is access to a PRI (Primary Rate
ISDN).
M
Braid Shield
In coaxial cable, a braid or mesh conductor,
made of copper or aluminum, that sur-
rounds the insulation and foil shield. The
braid helps protect the carrier wire from
electromagnetic and radio frequency
interference.
SEE ALSO
Cable, Coaxial
MBRI (Basic Rate Interface)
A BRI is an interface between a user and an
ISDN (Integrated Services Digital Network)
THE DOS BOOTSTRAP PROCESS
1. A program (the ROM-BIOS) in ROM (read-only -memory) executes. This program checks the hardware
components by doing a POST (power-on self-test).
2. The ROM-BIOS program loads and executes a program from the boot sector on a floppy or hard disk.
3. This boot sector program loads hidden files, which, in turn, load the basic device drivers for DOS (key-
board, disk, and display) and execute the DOS initialization code. Part of this initialization loads the DOS
kernel.
4. The DOS kernel builds various tables it will need, initializes device drivers, and executes instructions found
in CONFIG.SYS, if this file exists.
5. The DOS kernel loads COMMAND.COM, the DOS command processor.


Prev Page 114 Next

102
Bridge
switch. The BRI specifies two 64 kilobit per
second (kbps) B channels (for voice and
data) and one 16 kbps D channel (for cus-
tomer and call information).
This channel combination is sometimes
denoted as 2B+D. It can be compared with
PRI (Primary Rate Interface).
Access to a BRI is provided by a BRA
(basic rate access).
MBridge
The term bridge generally refers to a hard-
ware device that can pass packets from one
network to another. Bridges operate at the
OSI Reference Model's second lowest layer,
the data-link layer. A bridge makes the net-
works look like a single network to higher
level protocols or programs.
A bridge serves both as a medium (the
bridge part) and as a filter. It allows packets
from a node on one network to be sent to a
node on another network. At the same time,
the bridge discards any packets intended
for the originating network (rather than
passing these to the other network).
The terms bridge and router are often used
interchangeably. In fact, in older documen-
tation, Novell referred to its routers as
bridges. A router is a device that can send
packets to network segments on the way to
their destination. Unlike bridges, routers
operate at the network layer of the OSI Ref-
erence Model. However, bridges and routers
have come to take on some of each others'
properties. In fact, a brouter (from bridging
router) is a device that has the capabilities of
both a bridge and a router.
A bridge's capability to segment, or
divide, networks is one difference between a
bridge and a repeater. A repeater is a device
that moves all packets from one network
segment to another by regenerating, retim-
ing, and amplifying the electrical signals.
The main purpose of a repeater is to extend
the length of the network transmission
medium beyond the normal maximum cable
lengths.
A bridge is independent of, and therefore
can handle packets from, higher level proto-
cols. This means that different higher level
protocols can use the same bridge to send
messages to other networks.
To protocols at higher OSI layers (most
immediately, the network layer), the pres-
ence of a bridge is transparent. This means
that two networks connected by a bridge are
treated as part of the same logical network
by protocols such as Novell's IPX/SPX,
IBM's NetBIOS, or the widely used TCP/IP.
This transparency makes it possible to
access a logical network that is much larger
than the largest physical network allowed.
Because it operates at the data-link layer, a
bridge just checks the address information in
a packet to determine whether to pass the
packet on. Beyond that checking, a bridge
makes no changes to a packet.
A bridge sees each packet that is trans-
mitted on each of the networks the bridge
connects. If a packet from network A is
Bridges versus
Routers, Brouters, and Repeaters
Protocol Independence of Bridges
Packet Transmission


Prev Page 115 Next

Bridge
103
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
addressed to a local node (that is, to one in
network A), the bridge discards the packet
since the packet will be delivered internally
through the network. On the other hand, if
a packet from network A is addressed to a
remote node (on network B), the bridge
passes the packet over to network B. The
figure "A simple local bridge" shows how
a bridge can connect two networks.
The bridge greatly reduces traffic on both
networks by protecting each network from
the other's local messages. This makes each
of the smaller networks faster, more reliable,
and more secure, while retaining transparent
communication with the other network (or
networks).
When routing packets, a bridge uses only
node addresses; it does not take network
addresses into account. A node address is a
physical address, associated with a network
interface card (NIC), rather than with a par-
ticular network.
Bridges can be categorized by several differ-
ent features. The table "Bridge groupings"
summarizes the various categories.
Types of Bridges
BRIDGE GROUPINGS
FEATURE
GROUPING
Level
Operation
Location
Bridged distance
LLC (logical-link-control) layer
versus MAC (media-access-
control) layer
Transparent versus source
routing
Internal (card) versus external
(stand-alone)
Local versus remote
MAC-layer bridges operate at the media-
access control (MAC) sublayer, the lower
sublayer into which the IEEE divides the
data-link layer of the OSI Reference Model.
These bridges can connect only networks
using the same architecture (Ethernet to
Ethernet, Token Ring to Token Ring, and so
on), because the bridge expects to handle a
particular packet format, such as Ethernet
or ARCnet.
LLC-layer bridges operate at the upper
sublayer of the data-link layer, the logical
link-level control (LLC) sublayer. These
types of bridges can connect different archi-
tectures (such as Ethernet to Token Ring),
because these architectures use the same
LLC sublayer format, even if they use differ-
ent formats at the MAC sublayer.
Most older bridges are of the MAC-layer
type and can connect only same-architecture
networks; most newer products are of the
LLC-layer type and can connect dissimilar
architectures.
The manner in which a bridge routes pack-
ets depends largely on the architectures
involved. Bridges connecting Ethernet net-
works use transparent routing, a packet-
routing method in which the bridge deter-
mines a route. Transparent bridges deter-
mine "on the fly" where a packet belongs.
Such bridges learn and store the location of
each node, and then route packets accord-
ingly. A transparent bridge can carry out its
routing without explicit instruction or atten-
tion from the user. The bridge determines
LLC Layer versus MAC Layer Bridges
Transparent Routing versus Source Routing


Prev Page 116 Next

104
Bridge
A SIMPLE LOCAL BRIDGE
the locations of a node by looking it up in a
table the bridge has built.
In contrast, most bridges connecting
Token Ring networks use source routing.
This is a deterministic routing method in
which the source node must provide the
route as well as the destination for the
packet. The source node learns the available
routes through route discovery. The routing
information is inserted by the sender and
can be determined by sending a discovery
packet. This packet uses the spanning tree
algorithm to find the most efficient route to
the destination and reports this route to the
sender.
Source routing bridges determine an
explicit path to the destination node and
include this routing information in the


Prev Page 117 Next

Bridge
105
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
packet. Surprisingly, the requirements for
source routing capabilities are considerably
more complex than for transparent bridges.
Accordingly, source routing capabilities are
generally available as options for a bridge.
Although source routing requires more work Local versus Remote Bridges
to find the path initially, it is more efficient
once the path has been established because
there is no longer any reason for the bridge
to find a path.
According to the IEEE 802.3 specifica-
tions, all bridges should be capable of using
transparent routing. Some can also do
source routing. A bridge can distinguish
between the two approaches by checking the
packet being sent. Depending on the value of
a particular bit in the source address field, a
packet may include source-routing
information.
A bridge may be internal or external. An
internal bridge is on a card plugged into an
expansion slot in a server. The server is part
of both networks. An internal bridge gets its
power from the PC's bus. Internal bridges
generally include multiple types of connec-
tors. A special type of internal bridge is used
to connect to wide-area networks (WANs).
This type of bridge will have connectors for
modem or telephone connections, such as
D-shell or RJ-type connectors.
An external bridge is a stand-alone com-
ponent to which each network is connected
by cable. The external bridge is part of both
networks. An external bridge generally has
multiple connectors; for example, BNC for
coaxial cable (as in Ethernet or ARCnet net-
works); modular (RJ-xx) for twisted-pair
cable, and possibly DB-9 or DB-25 (for
serial connection to a modem). External
bridges need their own power supply and
they usually include a connector for access-
ing WANs.
A bridge may be local or remote. A local
bridge connects two networks in the same
geographical location, such as networks on
either side of the hall or on either side of an
office floor. Usually, these types of bridges
are added to break a large, busy network
into two smaller networks. This reduces net-
work traffic on each of the newly formed
networks.
By using the spanning tree algorithm
specified in the IEEE 802.1 standard, local
bridges can ensure that only a single path is
used to send a packet between a source and
a destination. If this path is not usable, the
algorithm can find an alternate path.
A remote bridge connects two networks
separated by considerable geographical dis-
tance, large enough to require a telecommu-
nications link. Remote bridges must be used
in pairs, with one at each end of the link, as
shown in the figure "A simple configuration
involving remote bridges."
A remote bridge connects to a local-area
network at one end and to a switching net-
work, such as one with an X.25 interface, at
the other end. Each remote bridge is con-
nected to a network at one port and to a net-
work cloud at another port. (A cloud is a
working concept that is used to indicate a
connection that is taken for granted, for pur-
poses of the discussion and whose details are
not specified.)
Internal versus External Bridges


Prev Page 118 Next

106
Bridge
The interfaces are likely to be different
at these two ports. For example, a remote
bridge may connect to an Ethernet network
at one port and to a serial interface (such as
RS-232) at the other. The cloud represents
the point-to-point link between the two
remote bridges.
A SIMPLE CONFIGURATION INVOLVING REMOTE BRIDGES


Prev Page 119 Next

Bridge
107
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
Remote bridges also need a protocol to
communicate with each other. For example,
if the remote bridges communicate over an
ISDN or an X.25 line, the bridge at each end
needs to be able to communicate using the
switched network (ISDN or X.25) protocol.
The throughput in a remote bridge is
likely to be limited by the long-distance
connection. At the local end, the bridge will
generally have the same nominal speed as
the network (10 Mbps for Ethernet, 4 or 16
Mbps for Token Ring, and so on). At the
remote end, the throughput will depend on
the type of connection. At this end, possible
speeds may run from a few kilobits per sec-
ond to several megabits per second.
A learning bridge is one that automatically
builds a table of node addresses, based on
the NICs the bridge finds on the network.
The bridge builds the table by using the
information broadcast when a new node
logs on and by checking on the source and
destination addresses as packets pass
through the bridge.
The performance of a learning bridge
improves over time as the bridge com-
pletes its table of node locations. Until it
knows the location of a node, the bridge
assumes the node is on the remote network
and so passes on the packets. The bridge is
constantly updating its table-adding new
addresses and dropping addresses that have
not been mentioned within a period of time.
In contrast, a static bridge is one that can-
not build its own address table. Instead, the
addresses must be entered by hand. Fortu-
nately, static bridges have all but disap-
peared. Just about all modern bridges are
learning bridges, since static bridges do not
meet IEEE 802.1 specifications.
Multiple bridges may be used to connect
several networks. Any one bridge connects
only two networks directly, but may connect
more than two networks indirectly. The
bridge is attached to each network by a port.
If there are multiple bridges, the bridges
communicate with each other and establish
a layout in order to find a spanning tree for
all the networks. A spanning tree is one that
includes paths to all nodes that can be
reached on the network but includes no
more paths than are necessary to completely
interconnect the nodes and networks
involved. Most important, a spanning tree
does not include any loops (closed paths)
which could trap a packet, thereby effec-
tively shutting down the network.
Because larger network clusters make
multiple paths possible, there is the danger
that the same message will get broadcast all
over the networks through multiple paths.
This will produce a great deal of extraneous
network traffic and can, in fact, bring down
the network. A closed path, or loop, among
the networks could be damaging because
it could start an unending packet-passing
process. The spanning-tree algorithm, speci-
fied in IEEE 802.1, is applied to provide a
path between every pair of accessible nodes
on the network and ensure that there are no
loops in the paths to be used by the bridge.
Although the spanning tree algorithm
ensures that the same packet won't take
multiple paths to the same destination, the
algorithm doesn't rule out the possibility of
Learning Bridges versus Static Bridges
Multiple Bridges and the
Spanning Tree Algorithm


Prev Page 120 Next

108
Bridge
WHAT TO LOOK FOR IN A BRIDGE
When you're investigating bridges, you'll want to get details about bridge features and capabilities. Vendors
should be able to provide both marketing and technical information about their products. Make sure
to get the technical information. The vendors' materials should provide information about at least the fol-
lowing:
I
Whether the bridge is local or remote.
I
Whether the bridge is internal or external.
I
Media and architecture supported for the local network; for example, twisted-pair Ethernet, 16 Mbps
Token Ring, or FDDI. It's a good idea to ask explicitly about your particular configuration and to get the
answer in writing.
I
If applicable, what interface the bridge supports for a remote connection. For example, it may support
RS-232, RS-422, V.35, T1, or DSx.
I
Number of ports.
I
Transmission speeds, both local and long distance, if applicable. The smaller of these values is the critical
one. Number of packets passed is generally a more useful figure than the actual bit-transfer rate.
I
Whether the bridge supports load balancing.
I
Whether the bridge can collect network performance data, such as number of packets received, for-
warded, and rejected, number of collisions, and errors during a transmission. Such network management
services may require additional software (which may cost several thousand dollars).
I
Price, which can range from a few hundred dollars to over $10,000.
When you're selecting a remote bridge, you need to worry about compatibility with the network and also
with the long-distance services that will be used. Keep in mind that you may need to budget for two remote
bridges if you're responsible for the networks at both ends of the connection.
For more specific and more advanced questions, such as about a bridge's compatibility with a particular net-
work configuration, you may need to talk to the bridge vendor's technical support staff. In many cases, the
network vendor (Novell, Banyan, and so on) will have a database of hardware that has been explicitly tested
with the vendor's networking products. Be forewarned that these vendors may want to charge you for
revealing this information.


Prev Page 121 Next

Broadband Transmission
109
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
multiple paths being used to transmit differ-
ent packets between the same source and
destination. Higher-end bridges include the
ability to do load balancing by distributing
traffic over more than one path between a
source and destination.
Recently, wireless bridges have become
available for limited-distance remote con-
nections. Remote bridges that use radio
waves can be up to 25 or 30 miles apart-
provided the terrain and weather allow it,
and provided the two bridges have direc-
tional antennas available. Remote bridges
using lasers can be up to about 3,500 feet
apart. Since focused signals must be sent in
both cases, such bridges must be within each
other's line of sight.
Wireless remote bridges are susceptible to
two kinds of interference:
Inward interference, which can occur
when another device is operating in the
same bandwidth and the two signals
interact with each other.
Outward interference, in which the device
under consideration is causing interfer-
ence in a different device.
BROADER CATEGOR Y
Internetwork Link
SEE ALSO
Brouter; Gateway; Repeater; Router;
Switch
M
Broadband Transmission
A broadband transmission is an analog
communication strategy in which multi-
ple communication channels are used
simultaneously. The data in a broadband
transmission is modulated into frequency
bands, or channels, and is transmitted in
these channels.
Guard bands, which are small bands of
unused frequencies, are allocated between
data channels. These provide a buffer
against interference due to signals from one
data channel drifting or leaking over into a
neighboring one. The figure "A broadband
transmission" shows how data channels and
guard bands are used.
For example, cable TV (CATV) uses
broadband transmission, with each channel
getting a 6 megahertz (MHz) bandwidth.
Broadband transmissions use coaxial or
fiber-optic cable and they can transmit voice,
data, or video.
A BROADBAND TRANSMISSION
1010
1100
1001
0001010100011010110
1010101010001010101
10100011101110
01010111110100
111000110101001
00010101000101


Prev Page 122 Next

110
Broadcast
When digital data is being transmitted, a
modem or other device demodulates the sig-
nals back into digital form at the receiving
end. A modem used for broadband trans-
missions needs two bands of at least 18
MHz bandwidth each: one band for sending
and the other for receiving.
MBroadcast
In a network transmission, sending a mes-
sage to all connected nodes. This is in con-
trast to a transmission that is targeted at a
single node. Most packet formats have a
special address value to indicate a packet
that is being broadcast. Compare broadcast
with multicast.
M
Broadcast Storm
In network traffic, a condition in which
packets are broadcast, received, and then
broadcast again by one or more of the
recipients. The effect of a broadcast storm is
to congest a network with redundant traffic.
Broadcast storms can arise, for example, in
bridged networks that contain loops (closed
paths).
MBroadcast Transmission
In an AppleTalk network that uses the
LocalTalk architecture and its LocalTalk
Link Access Protocol (LLAP), a transmission
sent to each node in the network. Compare
broadcast transmission with directed
transmission.
M
Brouter
A brouter (also known as a bridging router
or, less commonly, as a routing bridge) is a
device that combines the features of a bridge
and a router. A brouter can work at either
the data-link layer or the network layer.
Working as a bridge, a brouter is protocol
independent and can be used to filter local-
area network traffic. Working as a router, a
brouter is capable of routing packets across
networks.
BROADER CATEGORIES
Bridge; Internetwork Link; Router
MBrownout
A short-term decrease in voltage level, spe-
cifically when the voltage is more than 20
percent below the nominal RMS voltage.
Brownouts can occur when a piece of heavy
machinery is turned on and temporarily
drains the available power, or when every-
one feels the need to run their air condition-
ers at the same time.
SEE ALSO
Power Disturbance
MBrowser
A browser is a hypertext file reader. That is,
a browser is a program that can display
material containing links to other material
(perhaps located in other files), and can pro-
vide quick and easy access to the contents
associated with such links.


Prev Page 123 Next

BTAM (Basic Telecommunications Access Method)
111
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
Browsers may be text, graphics, or multi-
media based:
I A text-based, or line-oriented, browser
is unable to display anything but rudi-
mentary graphics, and is generally line-
oriented, but can still switch to any
material that is formatted in a suitable
manner for the browser. WWW and
Lynx are examples of such browsers.
Both are accessible on the Internet.
I Graphics browsers can handle both
text and graphics, require a mouse,
and generally have a much nicer dis-
play than line-oriented browsers. Cello M
and Mosaic are examples of graphics-
based browsers.
I Multimedia browsers can display
sound and video, in addition to the
capabilities of graphics browsers.
Mosaic is also a multimedia browser.
Variants of the mosaic browser are
available for several computing envi-
ronments. For example, XMosaic is a
browser for the X Window System.
HotJava, a recently announced
browser from Sun Microsystems is
generally regarded as taking browser
technology to a new level. HotJava can
handle multimedia material, includes
security capabilities, and offers an
object-oriented programming language M
for creating platform-independent
applications easily. Because HotJava
differs so drastically from existing
browsers, it remains to be seen how
quickly- or whether-HotJava
becomes widely used.
Forms-capable browsers allow users to
fill in information on forms or question-
naires. Most graphics-based browsers are
forms-capable.
Browsers have long been used in pro-
gramming environments-for example, in
the SmallTalk environment created at Xerox
PARC in the 1970s and 1980s. These read-
ers have really come into widespread use
with the growth of the World Wide Web
(WWW) on the Internet.
SEE ALSO
HotJava; Mosaic; WWW
BSD Socket Layer
In BSD UNIX, the layer that represents
the API (Application Program Interface)
between user applications and the network-
ing subsystem in the operating system
kernel.
M
BSD UNIX (Berkeley Software
Distribution UNIX)
A UNIX version implemented at the Univer-
sity of California, Berkeley. BSD UNIX
introduced several enhancements to AT&T's
original implementation, including virtual
memory, networking, and interprocess com-
munication support.
BTAM (Basic Telecommunications
Access Method)
An early access method for communications
between IBM mainframes and terminals.
BTAM is still used, but is largely obsolete
because it does not support IBM's SNA (Sys-
tems Network Architecture). ACF/VTAM


Prev Page 124 Next

112
Btrieve
has replaced BTAM as the method of choice
for remote communications with IBM
mainframes.
MBtrieve
In Novell's NetWare 3.0 and later, Btrieve is
a key-indexed record management program
that allows you to access, update, create,
delete, or save records from a database.
Btrieve is a program (actually several
programs) that can run in either of two ver-
sions: client- or server-based.
In addition to record-management capa-
bilities, Btrieve includes the following:
I Communications facilities, for
both local and remote communica-
tions between a program and a record
base. The Btrieve Message Routers
(that is, BROUTER.NLM and
BDROUTER.NLM) handle outgoing
requests; BSPXCOM handles incom-
ing requests from a remote source (a
workstation or another server).
I Requesters (DOS, OS/2, and so on),
which provide Btrieve access for appli-
cations running on workstations. The
requesters are: BREQUEST.EXE (for
DOS), BTRCALLS.DLL (for OS/2),
and WBTRCALL.DLL (for Windows).
I Utilities for setting up, monitoring,
and maintaining the record base,
among other things. These utilities are
mentioned briefly in the next section.
I Special data-protection measures for
dealing with the record base in case of
system failure. In addition to the stan-
dard ones such as record locking, data
protection measures include logging,
which records any changes made to
designated files so that the changes can
be undone later, if necessary. The roll
forward modules mentioned in the
next section provide the mechanism
for such corrections. Data protection
measures also include shadow paging,
in which page images are saved before
making any changes on the page.
Btrieve can back up files even while
they're in use by using continuous
operation.
I Support for NetWare Directory Ser-
vices (NDS), which are new with Net-
Ware 4.x. This support is available
only beginning with version 6.1 of
Btrieve.
I Security measures such as the ability to
encrypt and decrypt data and also the
ability to assign ownership to files.
I Memory management and caching
capabilities to help speed up access and
other operations.
Btrieve creates and maintains a key-
indexed record base (or database). A
key-indexed database is one in which keys,
or record fields, are used as the basis for cre-
ating an index, which is information that
guides access to a database.
A Btrieve record base uses a specially
defined data format, which is also supported
by database programs and other applica-
tions from third-party vendors.


Prev Page 125 Next

Btrieve
113
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
rstuvwxyz
The Btrieve programs are provided in Net-
Ware Loadable Modules (NLMs). The most
fundamental of these are BTRIEVE.NLM
and BSPXCOM.NLM.
BTRIEVE contains the Record Manager
program that does the work on the server.
This program performs disk I/O (input/out-
put) for Btrieve files on the server. This pro-
gram must be loaded on any server that has
Btrieve files.
BSPXCOM handles requests to the server
from any workstation or another remote
source. BSPXCOM must be loaded on any
server that needs to communicate with a
Btrieve requester program on a workstation.
Such a Btrieve requester must be loaded
on any workstation that needs to communi-
cate with a Btrieve record base. This pro-
gram relays requests from the user or from
an application to the Record Manager on
the appropriate server.
Other NLMs handle more specialized
duties. For example, BROUTER.NLM and
BDROUTER.NLM handle Btrieve-related
requests from a server to a remote server.
The figure "Relationships of Btrieve ele-
ments" shows how the various Btrieve
elements fit together.
Several Btrieve utilities provide the more
nitty-gritty services needed to handle the
record bases:
I BTRMON.NLM monitors Btrieve
activity on the server.
I BSETUP.NLM and BREBUILD.NLM
are used to change configurations and
Btrieve-Related Modules
RELATIONSHIPS OF BTRIEVE ELEMENTS


Prev Page 126 Next

114
BTU (Basic Transmission Unit)
to update Btrieve data files from ver-
sion 5.x to 6.x, respectively.
I BUTIL.NLM imports and exports
Btrieve data, and transfers data
between Btrieve files.
I BDIRECT.NLM provides support for
the NDS in NetWare 4.x. This NLM is
available only in Btrieve versions 6.1
and later.
I BROLLFWD.EXE (for DOS),
PBROLL.EXE (for OS/2), and
WBROLL.EXE (for Windows) are
the roll forward utilities. These are
used to restore a Btrieve file in case
of some type of system failure.
The server-based version runs the Btrieve
Record Manager on the server and a special
(operating system dependent) requester pro-
gram on the workstation. The Record Man-
ager handles the I/O for the database; the
requester handles the I/O between worksta-
tion and server.
The client-based version does all its pro-
cessing on the workstation, and makes I/O
calls (calls involving the record base)
through the workstation's operating system.
The client-based version is available only to
developers who want to create applications
that can use Btrieve data files.
If the calls are for the server's record base,
the Btrieve requester redirects the calls to the
server. The figure "A client and server using
Btrieve" shows this situation. Note that the
Btrieve requester is provided as part of a
server-based Btrieve implementation.
BROADER CATEGOR Y
NetWare
M
BTU (Basic Transmission Unit)
In IBM's SNA communications, an aggre-
gate block of one or more path information
units (PIUs) that all have the same destina-
tion. Several PIUs can be combined into a
single packet, even if they are not all part of
the same message. BTUs are created at the
path-control layer.
SEE ALSO
SNA (Systems Network Architecture)
M
Buffer, Fiber-Optic Cable
In fiber-optic cabling, a layer immediately
surrounding the cladding (which surrounds
the fiber core). The tighter this buffer is
wrapped around the cladding, the less
opportunity the cladding and core have to
move around in the cable.
SEE ALSO
Cable, Fiber-Optic
M
Buffer, Memory
In memory or storage applications, a buffer
is a temporary storage location that is gener-
ally used to hold intermediate values, or
other types of data, until they can be pro-
cessed. The storage may be allocated in ordi-
nary RAM (random-access memory), on a
hard disk, or in special memory registers
(such as on a UART chip, which is used for
serial communications).
A print buffer is one common example. A
spooler program saves a file to be printed in
the print buffer, and deals with the file as
Server- and Client-Based Btrieve


Prev Page 127 Next

Burned-In Address (BIA)
115
a
b
c
d
efghijklmno
p
q
r
s
t
u
v
w
x
y
z
CPU (central processing unit) availability
allows. Buffers provide faster access to
stored data.
Three types of buffer allocations are
distinguished:
File-cache buffer: Used to store disk-
allocation blocks temporarily.
Directory-cache buffer: Used to store the
DET (directory-entry table) blocks.
Packet-receive buffer: Used to hold
incoming packets until they can be
processed.
M
Buffered Repeater
In a network cabling scheme, a device that
can clean and boost signals before sending
them on. A buffered repeater can hold a
message temporarily for example, when
there is already a transmission on the
network.
SEE ALSO
Repeater
M
Burned-In Address (BIA)
SEE
BIA (Burned-In Address)
A CLIENT AND SER VER USING BTRIEVE


Prev Page 128 Next

116
Burstiness
M
Burstiness
In the CCITT recommendations for B-ISDN,
a measure of the distribution of data over
time. The definition for the term has not yet
been finalized. One definition being consid-
ered is the ratio between maximum, or peak,
and mean (average) bit rate.
M
Burst Mode
A high-speed transmission mode in which
the transmitter takes control of the commu-
nications channel temporarily, until its
transmission is complete. This mode is used
in internal communications, such as between
hard disk and bus, and also in communica-
tions between devices. The term is also used
to refer to the packet burst protocol in
NetWare.
MBurst Speed
The maximum speed at which a device can
operate without interruption, generally only
for short periods. This is in contrast to
throughput, which indicates the average
speed at which a device can operate under
ordinary conditions, such as when transmit-
ting or printing an entire file.
MBus
In computer hardware, a bus is a path
for electrical signals, generally between the
CPU (central processing unit) and attached
hardware. Buses differ in the number of bit
values they can carry at a time, in their
speed, and in their control mechanisms:
Bit values: In the PC world, 8-, 16-, and
32-bit data buses are common. On
workstations and larger machines, 64-
and 80-bit buses are common.
Speed: The speed of a bus depends on the
system clock. Bus speed is generally
measured in megahertz (MHz). The
IBM-PC bus has gone from a 4.77
MHz clock speed in the original PC to
66 MHz in today's high-end machines.
Other chips can support clock speeds
of over 100 MHz.
Control: Buses may be controlled through
interrupts or through polling.
In networking, bus refers to a logical and
physical network topology in which mes-
sages are broadcast along the main cable, so
that all nodes receive each transmission at
the same time. Standard Ethernet and cer-
tain ARCnet networks use a bus topology.
SEE ALSO
Topology, Bus
MBus Interface Board (BIB)
SEE
BIB (Bus Interface Board)
MBus Interface Unit (BIU)
SEE
BIU (Bus Interface Unit)


Prev Page 129 Next

Byzantine Failure/Byzantine Robustness
117
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
M
Bus Mastering
In general, bus mastering is a bus-access
method in which a card or device takes con-
trol of the bus in order to send data onto the
bus directly, without help from the CPU
(central processing unit). In a network,
the network interface card takes control
of the bus.
Generally, MCA (Microchannel Architec-
ture) and EISA (Extended Industry Standard
Architecture) machines support bus master-
ing, but ISA (Industry Standard Architec-
ture) machines do not. VL (VESA local) and
PCI (Peripheral Component Interconnect)
buses also support bus mastering.
Bus mastering can improve throughput
considerably, but only if the board and the
computer support the same bus-mastering
method, and if the bus mastering doesn't
conflict with the hard-disk controller.
Several types of transfer modes are possi-
ble with bus mastering, including burst
mode, streaming data mode, and data
duplexing. A particular bus-mastering
scheme may support some or all of these
modes.
M
Bypass
In telephony, a connection with an inter-
exchange carrier (IXC) that does not go
through a local exchange carrier.
MByte
A collection of-usually eight-bits (but
rarely worth a dollar anymore). A byte
generally represents a character or digit.
M
BYTE Information Exchange (BIX)
SEE
BIX (BYTE Information Exchange)
MByte-Sex
For a processor, byte-sex is a feature that
describes the order in which bytes are repre-
sented in a word. Processors may be little-
endian, big-endian, or bytesexual.
In little-endian representations, the low-
order byte in a word is stored at the lower
address. In big-endian processors or con-
texts, the high-order byte is stored first.
Bytesexual is a term used to describe a pro-
cess that is capable of using either little-
endian or big-endian representations for
information, depending on the value of a
flag bit.
SEE ALSO
Big-Endian; Little-Endian; Middle-Endian
M
Byzantine Failure/Byzantine
Robustness
In networking, a situation in which a node
fails by behaving incorrectly or improperly,
rather than by breaking down completely
and disappearing from the network. A net-
work that can keep working even if one or
more nodes is experiencing Byzantine failure
has Byzantine robustness.


Prev Page 130 Next



Prev Page 131 Next

CC


Prev Page 132 Next

120
Cable
C
M
Cable
It took about 100 years for cable to replace
the kite string as a medium for electrical
power, but the change was heartily wel-
comed, particularly by researchers. Cables
are currently the most popular medium for
transmitting information between nodes in
a network, although wireless transmission
schemes (radio, infrared, and microwave
communications) are becoming more widely
used.
In a network, the cabling scheme connects
nodes (or stations) and also gives the net-
work its characteristic shape (topology) and
features. Network cabling schemes distin-
guish between main and auxiliary cables.
The main cable provides the path and
defines the shape for the network; the auxil-
iary cables connect nodes to the main path
or to wiring centers that are connected to
the main path. Depending on the architec-
ture, the terminology for such cables differs.
For Ethernet networks, the main cable is
referred to as the trunk cable, and the auxil-
iary cables are called drop cables. Trunk
cable forms the backbone, or main cabling
scheme, of an Ethernet network. Because of
its role and location, trunk cable is some-
times called backbone cable. Drop cable
may be used to attach an individual node to
a network trunk cable. Nodes can also be
connected to the cable indirectly through a
connector or transceiver rather than with
drop cable. The different types of connectors
are discussed in a separate article.
IBM Token Ring networks distinguish
between the main ring path and patch
cables. In this context, patch cables attach
nodes (called lobes in Token Ring networks)
to wiring centers. The wiring centers are
called multistation attachment units (MAUs)
in such networks. The patch cables can also
attach to patch panels, which are, in turn,
connected to MAUs.
Four main types of cable are used in
networks:
I Coaxial cable, also called coax, which
can be thin or thick.
I Twisted-pair cable, which can be
shielded (STP) or unshielded (UTP).
I IBM cable, which is essentially
twisted-pair cable, but designed to
somewhat more stringent specifica-
tions by IBM. Several types are
defined, and they are used primarily
in IBM Token Ring networks.
I Fiber-optic cable, which can be single-
mode, multimode, or graded-index
multimode.
Coaxial, IBM, and twisted-pair cables
transmit electricity. Fiber-optic cables trans-
mit light signals. Each of the cable types is
subdivided into more specialized categories
and has its own design and specifications,
standards, advantages, and disadvantages.
Network Cabling Schemes
Ethernet Trunk and Drop Cables
IBM Token Ring
Cable Types


Prev Page 133 Next

Cable
121
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
Cable types differ in price, transmission
speed, and recommended transmission dis-
tance. For example, twisted-pair wiring is
currently the cheapest (and also the most
limited in performance). Fiber-optic cable is
more expensive but much faster and more
robust. Coaxial cable lies between these two
types on most performance and price
features.
This article discusses network cabling in
general. The specific cable types (coaxial,
twisted-pair, IBM, and fiber-optic) are cov-
ered in more detail in separate articles. In
addition to this cabling, there is a cable
infrastructure behind the walls, in shafts,
and under the ground. These cables are dis-
cussed under the headings Cable, Horizontal
and Cable, Backbone.
The different cable types have the following
components in common:
I A conductor to provide a medium for
the signal. The conductor might be a
copper wire or a glass tube.
I Insulation of some sort around the
conductor to help keep the signal in
and interference out.
I An outer sheath, or jacket, to
encase the cable elements. The jacket
keeps the cable components together,
and may also help protect the cable
components from water, pressure, or
other types of damage.
In addition to these common features,
particular types of cable have other compo-
nents. Coaxial cable has one or more shields
between the insulation and the jacket.
Twisted-pair cable has two conductor wires
twisted around each other. Fiber-optic cable
may include material to help protect the
fiber from pressure.
For electrical cable, the conductor is known
as the signal, or carrier, wire, and it may
consist of either solid or stranded wire. Solid
wire is a single thick strand of conductive
material, usually copper. Stranded wire con-
sists of many thin strands of conductive
material wound tightly together.
Signal wire is described in the following
terms:
I The wire's conductive material (for
example, copper)
I Whether the wire is stranded or solid
I The carrier wire's diameter, expressed
directly (for example, in inches, centi-
meters, or millimeters), or in terms of
the wire's gauge, as specified in the
AWG (American Wire Gauge) tables
(see the AWG article for a summary of
gauges)
The total diameter of the strand deter-
mines some of the wire's electrical proper-
ties, such as resistance and impedance.
These properties, in turn, help determine the
wire's performance.
For fiber-optic cable, the conductor is
known as the core. The core is a glass or
plastic tube that runs through the cable. The
diameter of this core is expressed in microns
(millionths of a meter).
Cable Components
Conductor


Prev Page 134 Next

122
Cable
The insulating layer keeps the transmission
medium's signal from escaping and also
helps to protect the signal from outside
interference. For electrical wires, the insula-
tion is usually made of a dielectric (noncon-
ductor), such as polyethylene. Some types of
coaxial cable have multiple protective layers Nonplenum Cable Jacket
around the signal wire.
For fiber-optic cable, the insulation is
known as cladding and is made of material
with a lower refraction index than the core's
material. The refraction index is a measure
that indicates the manner in which a mate-
rial will reflect light rays. The lower refrac-
tion index ensures that light bounces back
off the cladding and remains in the core.
The outer casing, or jacket, of the cable pro-
vides a shell that keeps the cable's elements
together. Two main classes of jacket are ple-
num and nonplenum. For certain environ-
ments, plenum cable is required by law. It
must be used when the cable is being run
"naked" (without being put in a conduit)
inside walls, and should probably be used
whenever possible.
Plenum jackets are made of nonflam-
mable fluoropolymers (such as Teflon or
Kynar). They are fire-resistant and do not
give off toxic fumes when burning. They are
also considerably more expensive (by a fac-
tor of 1.5 to 3) than cables with nonplenum
jackets. Studies have shown that cables with
plenum jackets have less signal loss than
nonplenum cables.
Plenum cable used for networks should
meet the NEC's CMP (National Electric
Insulation Layer
Plenum Cable Jacket
Code's communications plenum cable)
or CL2P (class 2 plenum cable) specifica-
tions. The cable should also be UL-listed for
UL-910, which subjects plenum cable to a
flammability test. The NEC and UL specifi-
cations are discussed in the Cable Standards
article.
Nonplenum cable uses less-expensive mate-
rial for jackets, so it is considerably less
expensive than cable with plenum jackets,
but it can be used only under restricted con-
ditions. Nonplenum cable jackets are made
of polyethylene (PE) or polyvinylchloride
(PVC), which will burn and give off toxic
fumes.
PVC cable used for networks should meet
the NEC's CMR (communications riser
cable) or CL2R (class 2 riser cable) specifica-
tions. The cable should also be UL-listed for
UL-1666, which subjects riser cable to a
flammability test. See the Cable Standards
article for a discussion of cable safety stan-
dards and performance levels.
Cable can be packaged in different ways,
depending on what it is being used for and
where it is located. For example, the IBM
cable topology specifies a flat cable for use
under carpets. Some fiber-optic trunks con-
tain thousands of fibers, each of which can
carry multiple messages.
The following types of cable packaging
are available:
Simplex cable: One cable within one
jacket, which is the default configu-
ration. The term is used mainly for
Cable Packaging


Prev Page 135 Next

Cable
123
a
b
c
d
e
f
g
h
ij
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
fiber-optic cable to indicate that the
jacket contains only a single fiber.
Duplex cable: Two cables, or fibers,
within a single jacket. In fiber-optic
cable, this is a common arrangement.
One fiber is used to transmit in each
direction.
Multifiber cable: Multiple cables, or
fibers, within a single jacket. For fiber-
optic cable, a single jacket may contain
thousands of fibers; for electrical cable,
the jacket will contain at most a few
dozen cables.
Cable is described in terms of the size and
makeup of its components, as well as in
terms of its performance. For example, elec-
trical cable specifications include the gauge,
or diameter, of the signal wire.
The cable's electrical and physical proper-
ties determine the performance you can
expect and the range of conditions under
which you can use the cable. Cables differ in
the electrical properties (signal loss, imped-
ance, and so on) they offer. The table "Cable
Properties" lists some of the features that
distinguish cables.
Cable Properties
CABLE PROPER TIES
PROPERTY
MEASUREMENT
OR DESCRIPTION
COMMENT
Size
Conductor
wire diameter
Core fiber diameter
Wire insulation
diameter
Cladding diameter
Wire shield
diameter
Jacket diameter
Millimeters (mm), inches
(in), or gauge (AWG)
Microns
Millimeters or inches
Microns
Millimeters, inches,
or gauge
Millimeters or inches
For stranded wire, this represents the total diameter of the
entire cluster of strands.
Some core diameters have desirable properties in terms of
the paths certain wavelengths of light take in the core. For
example, diameters of 62.5 and 100 microns for multimode
fiber and of under 10 microns for single-mode fiber are
common.
The diameter of the cable's insulaton layer is needed to
calculate certain electrical properties of a cable.
The cladding diameter varies much less than the core
diameter, partly because the cladding helps to make the
fiber easier to package if the cladding is of an approximately
constant size.
The diameter of the jacket can be important when installing
the cable because it may determine space requirements.


Prev Page 136 Next

124
Cable
PROPERTY
MEASUREMENT
OR DESCRIPTION
COMMENT
Composition
Conductor wire
composition
Wire insulation
composition
Shield composition
Jacket composition
Electrical
Properties
DCR (DC
Resistance)
Shield DCR
Impedance
Capacitance
Attenuation
Materials; solid vs.
stranded (# of strands)
Materials
Materials; % area cov-
ered by shield mesh
Materials; plenum vs.
nonplenum
Ohms ( ) per distance
(100 or 1000 feet)
Ohms ( ) per distance
(100 or 1000 feet)
Ohms
Picofarads per foot
(pF/ft)
Maximum decibels
per distance at a given
frequency; common dis-
tances include 100 feet,
1000 feet, and 1 kilome-
ter, e.g., dB/1000 ft at
5 MHz
Conductor wires may be solid or stranded, or of different
types of conductive material (usually copper alone or in
some variant). If the wire is stranded, the specifications
should note the number of strands.
For coaxial cable only, shield composition refers to the
makeup of the protective shield around the conductive
wire.
Refers to the DC resistance for the conductor wire.
Refers to the DC resistance for the shield.
The measure of a wire's resistance to electrical current,
which helps determine the wire's attenuation properties.
Most networks use cable with a characteristic impedance
level. There are devices for connecting cable segments that
have diffferent impedances.
The measure of the cable's ability to store up electrical
charge or voltage. This charge storage distorts a signal as
it travels along its course; the lower the capacitance
the better.
The measure of the signal loss over distance. Data sheets
may include several attenuation values for different frequen-
cies. This distinction can be imporant because attenuation
of an electrical signal increases with signal frequency.


Prev Page 137 Next

Cable
125
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
PROPERTY
MEASUREMENT
OR DESCRIPTION
COMMENT
Crosstalk (NEXT)
Velocity of
Propagation
Other Properties
Weight
Maximum recom-
mended cable
segment range
Bandwidth
Price
Performance/Safety
Ratings
Minimum decibels per
distance (1000 or 100
feet) (dB/distance)
% (values should be
about 60%; preferably
above 80%)
Unit weight per distance
(oz/ft; gm/meter)
Distance (feet, meters,
or kilometers)
Megahertz (MHz) or
megabits per second
(Mbps)
Dollars per distance
(100 or 1000 feet)
NEC CL2, CMP, and
CMR; EIA/TIA-568
Categories 1-5; UL
Levels 1-5; ETL ratings
NEXT (near-end crosstalk) is a common measure of inter-
ference by a signal from a neighboring cable or circuit. The
higher the decibel value, the less crosstalk.
Specifies the maximum signal speed along the wire, as a
proportion of the theoretical maximum (the speed of light).
See the Cable Standards article for information about these
cable safety standards.
You can obtain the specifications for a
specific type of cable from the cable manu-
facturer or vendor. The table "Cable Com-
ponent Abbreviations" lists some common
abbreviations used in cable specifications or
data sheets.
Cables are good media for signals, but they
are not perfect. Ideally, the signal at the end
of a stretch of cable should be as loud and
clear as at the beginning. Unfortunately, this
will not be true.
Any transmission consists of signal and
noise components. Even a digital signal
degrades when transmitted over a wire or
through an open medium. This is because
the binary information must be converted to
electrical form for transmission, and because
the shape of the electrical signal changes
over distance.
Factors Affecting Cable Performance


Prev Page 138 Next

126
Cable
CABLE COMPONENT ABBREVIATIONS
ABBREVIATION
FEATURE
COMPONENT(S)
AD
AL
ALS
AWG
BC
CCAL
CCS
FEP
FFEP
FP
K
PE
PVC
PVDF
SC
TC
x%
#cond
Air dielectric
Aluminum braid
Aluminum sheath
American Wire Gauge (AWG) value for wire
Bare copper braid
Copper-clad aluminum
Copper-covered steel
Fluorinated ethylene propylene (Teflon)
Foamed fluorinated ethylene propylene (Teflon)
Foamed polyethylene
Kynar/polyvinylidene fluoride (plenum)
Polyethylene (solid)
Polyvinylchloride
Generic polyvinylidene fluoride (plenum)
Silvered copper braid
Tinned copper braid
Percentage of surface area covered by braid
Number of conductors
Insulation
Shield
Shield
Carrier wire
Carrier wire; shield
Carrier wire
Carrier wire
Insulation; jacket
Insulation
Insulation
Jacket
Insulation; jacket
Jacket
Jacket
Carrier wire; shield
Carrier wire; shield
Shield
Carrier wire
Signal quality degrades for several rea-
sons, including attenuation, crosstalk, and
impedance.
Attenuation is the decrease in signal
strength, measured in decibels (dB) per 100
feet or per kilometer. Such loss happens as
the signal travels over the wire. Attenuation
occurs more quickly at higher frequencies
and when the cable's resistance is higher.
In networking environments, repeaters
are responsible for cleaning and boosting a
signal before passing it on. Many devices are
repeaters without explicitly saying so. For
example, each node in a token ring network
acts as a repeater. Since attenuation is sensi-
tive to frequency, some situations require the
Attenuation


Prev Page 139 Next

Cable
127
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
use of equalizers to boost different
frequency signals the appropriate amount.
Crosstalk is interference in the form of a sig-
nal from a neighboring cable or circuit; for
example, signals on different pairs of twisted
wire in a twisted-pair cable may interfere
with each other. A commonly used measure
of this interference in twisted-pair cable
is near-end crosstalk (NEXT), which is
represented in decibels. The higher the deci-
bel value, the less crosstalk and the better
the cable.
Additional shielding between the carrier
wire and the outside world is the most com-
mon way to decrease the effects of crosstalk.
Impedance, which is a measure of electrical
resistance, is not directly a factor in a cable's
performance. However, impedance can
become a factor if it has different levels
at different locations in a network. In order
to minimize the disruptive effects of differ-
ent impedances in a network, special de-
vices, called baluns, are used to equalize
impedance at the connection (at the balun
location).
Impedance does reflect performance indi-
rectly, however. In general, the higher the
impedance, the higher the resistance, and
the higher the resistance, the greater the
attenuation at higher frequencies.
Cables are used to meet all sorts of power
and signaling requirements. The demands
made on a cable depend on the location
in which the cable is used and the function
for which the cable is intended. These
demands, in turn, determine the features
a cable should have.
Here are a few examples of considerations
involving the cable's function and location:
I
Cable designed to run over long
distances, such as between floors or
buildings, should be robust against
environmental factors (moisture, tem-
perature changes, and so on). This may
require extra jackets or jackets made
with a special material. Fiber-optic
cable performs well, even over dis-
tances much longer than a floor or a
building.
I Cable that must run around corners
should bend easily, and the cable's
properties and performance should not
be affected by the bending. For several
reasons, twisted-pair cable is probably
the best cable for such a situation
(assuming it makes sense within the
rest of the wiring scheme). Of course,
another way to get around a corner is
by using a connector; however, con-
nectors may introduce signal-loss
problems.
I Cable that must run through areas in
which powerful engines or motors are
operating (or worse, being turned on
and off at random intervals) must be
able to withstand magnetic interfer-
ence. Large equipment gives off strong
magnetic fields, which can interfere
Crosstalk
Impedance
Selecting Cable
Function and Location


Prev Page 140 Next

128
Cable
with and disrupt nearby signals. In
commercial and residential settings,
this can be a problem with cable that is
run, for example, through the elevator Main Cable Selection Factors
shaft. Because it is not affected by such
electrical or magnetic fluctuations,
fiber-optic cable is the best choice in
machinery-intensive environments.
I If you need to run lots of cables
through a limited area, cable weight
can become a factor, particularly if all
that cable will be running in the ceiling
above you. In general, fiber-optic and
twisted-pair cable tend to be lightest.
I Cables being installed in barely acces-
sible locations must be particularly
reliable, and they should probably be
laid with backup cable during the ini-
tial installation. Some consultants and
mavens advise laying a second cable
whenever you are installing cable, on
the assumption that the installation is
much more expensive than the cable
and that installation costs for the sec-
ond cable add only marginally to the
total cost. Generally, the suggestion is
to make at least the second cable opti-
cal fiber.
I Cables that need to interface with
other worlds (for example, with a
mainframe network or a different
electrical or optical system) may need
special properties or adapters. For
example, UTP cable in a Token Ring
network needs a media filter between
the cable and the MAU to which the
cable is attached. The kinds of cable
required will depend on the details of
the environments and the transition
between them.
Along with the function and location con-
siderations, cable selections are determined
by a combination of factors, including the
following:
I The type of network you plan to create
(Ethernet, Token Ring, or another
type). While it is possible to use just
about any type of cable in any type of
network, certain cable types have been
more closely associated with particular
network types. For example, Token
Ring networks use twisted-pair cable.
I The amount of money you have avail-
able for the network. Keep in mind
that cable installation can be an expen-
sive part of the network costs.
I Whatever cabling resources are already
available (and usable). You will almost
certainly have available wiring that
could conceivably be used for a net-
work. It is almost equally certain,
however, that at least some of that
wire is defective or is not up to the
requirements for your network.
I Building or other safety codes and
regulations.
You can get cable with or without connec-
tors at either end. Both connected and bulk
cable have advantages and drawbacks.
Whether connected or bulk cable is better
depends on how you are going to use it.
Connected versus Bulk Cable


Prev Page 141 Next

Cable
129
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
You have much more flexibility to cut or
reroute with bulk cable, because you are not
restricted to a precut cable segment. On the
other hand, you (or someone you trust) will
need to attach the connectors. This requires
special tools and involves stripping the end
of the cable and crimping the connector to
the bare wire.
Cable prices depend on factors such as the
following:
I Type of cable (coaxial, twisted-pair,
fiber optic). In general, fiber-optic
cable is the most expensive but the
price is dropping rapidly. Fiber-optic
cable is followed closely by thick coax-
ial cable. STP and thin coaxial follow
in roughly that order, but with consid-
erable overlap in prices. UTP is the
least expensive type of cable.
I Whether cable comes in bulk or with
connectors at either end. While price is
an issue, this question will be answered
mainly by your needs for the cable.
I Whether the cable is plenum or non-
plenum. Plenum versions can cost
from 1.5 to 3 times as much as the
nonplenum version.
Cable prices change, so do not be sur-
prised to find considerable variation in
prices when you start getting quotes.
UTP cable is grouped into voice- and
data-grade. Most telephone wire is just
voice-grade. Prices for data-grade UTP cable
are a few cents higher per foot.
Cable Prices
Installation tools for handling cables include
wire strippers, dies, and crimping tools for
attaching connectors to the end of a stretch
of bulk cable. Such tools are often included
in adapter kits, which are configured for
building particular types of cable (for
example, coaxial cable or cable for RS-232
connections). Depending on how compre-
hensive the toolkit is, expect to pay any-
where from about $30 to $500.
Testing tools for cables include a whole
range of line scanners and monitors. The
simplest of these can tell you whether there
is any electrical activity between one loca-
tion in a network (or a cable installation)
and another. The most sophisticated can do
just about everything except tell you where
you bought the cable.
The top-of-the-line scanners can test any
kind of copper-based cable not only for
faults, but also for performance specifica-
tions (NEXT, attenuation, and so on). These
types of scanners know about the electrical
requirements of the most popular network
architectures (such as Ethernet/802.3 and
Token Ring) and are capable of finding
faults or deviations from specifications at
just about any location on the network.
Of course, you will pay several thousand
dollars for this capability.
Many companies sell both electrical and
fiber-optic cable, as well as connectors,
installation, and testing tools. Some vendors
specialize in fiber-optic products, others in
copper-based products, and still others offer
both.
Cabling Tools
Cable Vendors and Resources


Prev Page 142 Next

130
Cable
CABLE TIPS
Here are some tips on purchasing and installing cabling:
I
Cables have quite a few properties that should be considered in making decisions. You can find information
about these cable properties in cable specifications or data sheets, which are available from cable vendors.
I
In general, cable that meets military specifications (MIL-SPECS) is designed to more stringent require-
ments, and so is a good choice for networks. This is even more true for connectors, because the military
specifications insist on durable and reliable connectors. (Connectors are particularly prone to shoddy
construction.)
I
Fiber-optic connectors are especially tricky to attach because fiber optics has such exact alignment
requirements. It's probably worth your while to let a professional attach these connectors.
I
When you're ordering cable, make sure it's clear whether you want cable with connectors or "raw" (bulk)
cable.
I
Make sure the cable is good quality. Otherwise, you'll have trouble after a while, as the insulation within
and outside the cable breaks down.
I
Test cable both before and after installing it.
I
While present needs are obviously the major determinant of cabling decisions, future plans should also be
taken into consideration. In general, at least consider installing cable one level more powerful than you
think you'll need.
I
When adding cable to an existing cabling system, find out exactly what kind of cable is already in place. The
safest thing is to get the actual part and specification information from the cable jacket, then order exactly
that from the same distributor (or a certified equivalent from a different manufacturer).
I
Before adding to existing cable, test it as thoroughly as possible. If the cable seems likely to have a major
breakdown within a few months, it's almost certainly better to replace it now.
I
Protect the cable as much as possible. Such measures should include protecting the cable from tempera-
ture or moisture changes, which can cause the cable to crack or melt.
I
Support the cable as much as possible, so that a hanging cable doesn't stretch because the cable's own
weight is pulling it downward.
I
Velcro cable ties can help make things neater, by enabling you to collect multiple loose wires into a single
cluster. The Rip-Tie Company in San Francisco is one vendor that offers these neatness aids.


Prev Page 143 Next

Cable, Backbone
131
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
When you are ready to start looking for
cabling and other components, it will be
worthwhile getting the cabling guides and
catalogs from several vendors. The guides
offer useful general-purpose hints and guide-
lines for selecting and installing cable.
Here are some cable vendors and their
telephone numbers:
AMP Incorporated: (800) 522-6752;
(717) 564-0100
Andrew Corporation: (800) 328-2696;
Fax (708) 349-5673
Berk-Tek: (800) 237-5835
Black Box Corporation: (800) 552-6816;
(412) 746-5500
Comm/Scope: (800) 982-1708;
(704) 324-2200; Fax (704) 459-5099
CSP (Computer System Products):
(800) 422-2537; (612) 476-6866;
Fax (612) 476-6966
FIS (Fiber Instrument Sales):
(800) 445-2901; (315) 736-2206;
Fax (315) 736-2285
Jensen Tools: (800) 426-1194;
(602) 968-6231; Fax (800) 366-9662
Trompeter Electronics: (800) 982-2639;
(818) 707-2020; Fax (818) 706-1040
SEE ALSO
Cable, Backbone; Cable, Coaxial; Cable,
Fiber-Optic; Cable, Horizontal; Cable,
IBM; Connector; Connector, Fiber-Optic
MCable, Adapter
Cable used to connect a Token Ring
network interface card (NIC) to a hub
or multistation access unit (MAU). IBM
Type 1 and Type 6 cable can be used for this
purpose. The IBM cables have a DB-9 or
DB-25 connector at the NIC end and an
IBM data connector at the MAU end.
MCable, Backbone
Backbone cable refers to the cable that
forms the main trunk, or backbone, of a
network, particularly an Ethernet network.
Individual nodes and other devices may be
connected to this cable using special adapt-
ers (such as transceivers) and a separate
stretch of cable (called the drop cable in an
Ethernet network) to the node.
More generally, backbone cable is defined
by the EIA/TIA-568 committee as any
"behind the scenes" cable-cable running
behind walls, in shafts, or under the
ground-that is not classified as horizontal
cable. (Horizontal cable is defined by the
EIA/TIA-568 committee as any cable that
goes from a wiring closet, or distribution
frame, to the wall outlet in the work area.)
This includes cable used to connect wiring
closets and equipment rooms.
The EIA/TIA-568 recognizes four
main types of backbone cable, and several
optional variants. These types are listed in
the table "EIA/TIA-568 Main and Optional
Types of Backbone Cable."


Prev Page 144 Next

132
Cable, Broadcast-Oriented
COMPARE
Cable, Horizontal
SEE ALSO
Cable
M
Cable, Broadcast-Oriented
Cable that is designed to carry video signals
sent from one location in the network,
known as the head-end. This type of cable is
generally designed for one-way communica-
tion, which makes it of limited value for use
as network cable.
M
Cable, Category x
A five-level rating system for telecommuni-
cations wiring, specified in the EIA/TIA-568
documents. These describe minimum perfor-
mance capabilities for unshielded twisted-
pair cable.
SEE
Cable Standards
M
Cable, CATV (Community Antenna
Television, or Cable Television)
Wiring used for the transmission of cable
television signals. CATV is broadband
coaxial cable and is generally wired for
one-directional transmission; that is, from
the cable station, or a head-end, to the
consumer. If the CATV cable is not one-
directional, it may be possible to use it
for network cabling.
MCable, Coaxial
Coaxial cable, often called coax, is used for
data transmissions. This cable's remarkably
stable electrical properties at frequencies
below 1 GHz (gigahertz) makes the cable
popular for cable television (CATV) trans-
missions and for creating local-area net-
works (LANs). Telephone company
switching offices also use coaxial cable to
route long-distance calls. The figure "Con-
text and properties of coaxial cable" sum-
marizes the features of this type of cable.
EIA/TIA-568 MAIN AND OPTIONAL TYPES OF BACKBONE CABLE
CABLE TYPE
MAIN
OPTIONAL
UTP
STP
Coaxial
Optical fiber
100-ohm, multipair UTP cable, to be used for
voice-grade communications only
150-ohm STP cable, such as that defined in the
IBM Cable System (ICS)
50-ohm thick coaxial cable, such as the cable
used in thick Ethernet networks
62.5/125-micron (step- or graded-index)
multimode optical fiber
100-ohm STP cable
75-ohm (broadband) coaxial cable,
such as CATV cable
Single-mode optical fiber


Prev Page 145 Next

Cable, Coaxial
133
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
rs
t
u
v
w
x
y
z
A coaxial cable consists of the following
layers (moving outward from the center):
Carrier wire: A conductor wire (the car-
rier, or signal, wire) is in the center.
This wire is made of (or contains)
copper and may be solid or stranded.
There are restrictions regarding the
wire composition for certain network
configurations. The diameter of the
signal wire is one factor in determining
the attenuation (loss) of the signal over
distance. The number of strands in a
multistrand conductor also affects the
attenuation.
Coaxial Cable Components
CONTEXT AND PROPER TIES OF COAXIAL CABLE
Context
Cable
Electrical

Twisted-Pair

Coaxial
Optical

Fiber-Optic
Stable and predictable electrical properties
At least one shield around conductor wire
Subject to electromagnetic interference
Variable impedance levels
Thin and thick varieties
Broadband and baseband varieties
Thin coaxial uses BNC/TNC connectors; thick coaxial uses N-series connectors
Twinaxial runs two cables within a single jacket
Triaxial and quadrax have extra shielding for special uses
Ethernet networks
ARCnet networks
Cable TV lines
Video cable
IBM mainframe and midrange-based networks (twinaxial)
Telephone switching offices
Coaxial Uses
Coaxial Properties


Prev Page 146 Next

134
Cable, Coaxial
Insulation: An insulation layer consists
of a dielectric (nonconductor) around
the carrier wire. This dielectric is usu-
ally made of some form of polyethyl-
ene or Teflon.
Foil shield: A thin foil shield around the
dielectric. This shield usually consists
of aluminum bonded to both sides of a
tape. Not all coaxial cables have foil
shielding; some have two foil shield
layers, interspersed with braid shield
layers.
Braid shield: A braid, or mesh, conductor,
made of copper or aluminum, that sur-
rounds the insulation and foil shield.
This conductor can serve as the ground
for the carrier wire. Together with the
insulation and any foil shield, the braid
shield protects the carrier wire from
electromagnetic interference (EMI) and
radio frequency interference (RFI). The
braid and foil shields provide good
protection against electrical interfer-
ence, but only moderate protection
against magnetic interference.
Jacket: An outer cover that can be either
plenum (made of Teflon or Kynar) or
nonplenum (made of polyethylene
or polyvinylchloride).
The figure "A coaxial cable has five lay-
ers" shows the makeup of a coaxial cable.
The layers surrounding the carrier wire also
help prevent signal loss due to radiation
from the carrier wire. The signal and shield
wires are concentric, or co-axial, hence the
name.
The main features that affect the perfor-
mance of coaxial cable are its composition,
width, and impedance.
The carrier wire's composition determines
how good a conductor the cable will be.
Copper is among the best materials for this
purpose. The IEEE specifies stranded copper
carrier wire with tin coating for thin coaxial,
and solid copper carrier wire for thick
coaxial.
A COAXIAL CABLE
HAS FIVE LAYERS
Coaxial Cable Performance


Prev Page 147 Next

Cable, Coaxial
135
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
Cable width helps determine the electrical
demands that can be made on the cable. In
general, thick coaxial can support a much
higher level of electrical activity than thin
coaxial.
Impedance is a measure of opposition
to the flow of alternating current. The prop-
erties of the dielectric between the carrier
wire and the braid help determine the cable's
impedance. Each type of network archi-
tecture uses cable with a characteristic
impedance.
Impedance helps determine the cable's
electrical properties and also limits the con-
texts in which the cable can be used. For
example, Ethernet and ARCnet architectures
can both use thin coaxial cable, but they
have different impedances; therefore, Ether-
net and ARCnet cables are not compatible.
In networks, the impedances range from
50 ohms (for an Ethernet architecture) to
93 ohms (for an ARCnet architecture).
A segment of coaxial cable has an end
connector at each end. The cable is
attached through these end connectors
to a T-connector, a barrel connector, another
end connector, or to a terminator. Through
these connectors, another cable or a hard-
ware device is attached to the coaxial cable.
In addition to their function, connectors
differ in their attachment mechanism and
components. For example, BNC connectors
join two components by plugging them
together and then turning the components to
click the connection into place. Different
size coaxial cable requires a different-sized
connector.
Coaxial Cable Connectors
For coaxial cable, the following types of
connectors are available:
I A BNC (bayonet nut connector) is
used for thin coaxial cable.
I The N-series connectors are used for
thick coaxial cable.
I
A TNC (threaded nut connector) may
be used in the same situations as a
BNC, provided that the other connec-
tor is also using TNC.
Connectors for coaxial cable should be
plated with silver, not tin. This improves the
contact and the durability of the connector.
Descriptively, coaxial cable is grouped
mainly into thin and thick varieties. Thin
coaxial cable is 3/16-inch in diameter and
is used for various network architectures,
including thin Ethernet (also known as
10Base2 or CheaperNet) and ARCnet.
When using this configuration, drop cables
are not allowed. Instead, the T-connector
must be connected directly to the network
interface card (NIC). This means the NIC
must have an on-board transceiver, known
as a medium attachment unit (MAU) in the
IEEE 802.3 standard.
Thick coaxial cable is 3/8-inch in diame-
ter. It is used for thick Ethernet (also known
as 10Base5 or ThickNet) networks, cable
TV (CATV), and other connections. Thick
coaxial is expensive and is notoriously diffi-
cult to install and work with. It is more
likely to be inherited than selected for use
in a network.
Thin versus Thick Coaxial


Prev Page 148 Next

136
Cable, Coaxial
Other descriptions of coaxial cable are
based on the contents of the cable, rather
than its size, as follows:
Twinaxial: Also known simply as twinax,
this coaxial cable has two carrier
wires, each with its own dielectric, or
insulation, layer. The wires are gener-
ally twisted around each other, which
helps reduce magnetic interference,
and are surrounded by a shield and a
jacket whose properties run the same
gamut as for ordinary coaxial cable.
This type of cable is used in IBM and
AppleTalk networks. For example,
twinaxial cable is used to connect IBM
5250 terminals to System/36 or AS/
400 computers.
Triaxial: Also known simply as triax, this
coaxial cable has extra shielding: an
inner braid surrounded by an inner
(nonplenum) jacket, surrounded by
an outer copper braid. This outer braid
is, in turn, surrounded by the outer
jacket. The extra shielding makes a big
difference because of the grounding
and improved protection.
Quadrax: This cable is a hybrid of triax-
ial and twinaxial cable. Quadrax has
the extra carrier wire with dielectric,
and also has the extra shielding of
triaxial.
Quad shield: This cable has four layers of
shielding: alternating layers of foil and
braid shields. Quad shield cable is used
in situations where heavy electrical
Cable Content Descriptions
interference can occur; for example, in
industrial settings.
Functionally, coaxial cable is grouped into
baseband and broadband varieties.
Baseband coaxial cable has one channel
over which a single digital message can be
sent, at speeds of up to 80 megabits per
second (Mbps). Thin coaxial is used for
baseband cable.
Broadband coaxial cable can carry sev-
eral analog signals (at different frequencies)
simultaneously. Each of these signals can be
a different message or a different type of
information. Thick coaxial cable can be
used for broadband transmissions in a
network.
Broadband coaxial can use a single
cable or multiple cables. In single-cable
broadband coaxial, frequencies are split;
for example, into 6 megahertz (MHz) chan-
nels for each station. Some channels are
allocated for bidirectional communication.
Dual-cable broadband coaxial uses one
cable for sending and one for receiving data;
each cable has multiple channels.
Note that broadband coaxial requires
much more planning than baseband coaxial.
For example, a broadband setup will prob-
ably need amplifiers for dealing with the
different broadband signals.
The following designations are used for
coaxial cable used in networks. These are
just a few of the available coaxial cable
types.
Baseband versus Broadband Cable
Coaxial Cable Designations


Prev Page 149 Next

Cable, Coaxial
137
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
RG-6: Used as a drop cable for CATV
transmissions. It has 75 ohms imped-
ance, is a broadband cable, and is
often quad-shielded.
RG-8: Used for thick Ethernet. It has 50
ohms impedance. The thick Ethernet
configuration requires other cable and
a MAU (transceiver). The other cable
required is a twisted-pair drop cable to
the NIC. The drop cable off RG-8
cable uses a 15-pin DIX (or AUI) con-
nector. RG-8 is also known as N-Series
Ethernet cable.
RG-11: Used for the main CATV trunk.
It has 75 ohms impedance and is a
broadband cable. This cable is often
quad shielded (with foil/braid/foil/
braid around the signal wire and
dielectric) to protect the signal wire
under even the worst operating
conditions.
RG-58: Used for thin Ethernet. It has
50 ohms impedance and uses a BNC
connector.
RG-59: Used for ARCnet. It has 75 ohms
impedance and uses BNC connectors.
This type of cable is used for broad-
band connections and also by cable
companies to connect the cable net-
work to an individual household.
RG-62: Used for ARCnet. It has 93 ohms
impedance and uses BNC connectors.
This cable is also used to connect ter-
minals to terminal controllers in IBM's
3270 system configurations.
Coaxial cable has the following advantages
over other types of cable that might be
used for a network. The advantages are gen-
eral and may not apply in a particular situa-
tion. Note also that advantages change or
disappear over time, as technology advances
and products improve.
I
Broadband coaxial can be used to
transmit voice, data, and even video.
I The cable is relatively easy to install.
I Coaxial cable is reasonably priced
compared with other cable types.
Coaxial cable has the following disadvan-
tages when used for a network:
I It is easily damaged and sometimes
difficult to work with, especially in
the case of thick coaxial.
I Coaxial is more difficult to work with
than twisted-pair cable.
I This type of cable cannot be used with
token ring network architectures.
I
Thick coaxial can be expensive to
install, especially if it needs to be
pulled through existing cable conduits.
I Connectors can be expensive.
I Baseband coaxial cannot carry inte-
grated voice, data, and video signals.
Advantages of Coaxial Cable
Disadvantages of Coaxial Cable


Prev Page 150 Next

138
Cable, Data-Grade
Almost all cable testers can deal with coaxial
cable. (See the Cable article for a discussion
of the tools used for cable testing.) For more
specialized tasks requiring tools, such as
crimpers and dies for attaching connectors
to cable, you will need versions specifically
designed for coaxial cable.
When in doubt, of course, ask the vendor
explicitly whether a particular tool will
work with coaxial cable.
SEE ALSO
Cable; Cable, Fiber-Optic; Cable,
Twisted-Pair; Connector
M
Cable, Data-Grade
Twisted-pair cable of sufficiently high qual-
ity to use for data transmission. In contrast,
voice-grade cable is more susceptible to
interference and signal distortion. In the
EIA/TIA-568 cable specifications, categories
2 through 5 are data-grade cable.
SEE ALSO
Cable, Twisted-Pair
M
Cable, Distribution
In broadband networks, a term for cable
used over intermediate distances (up to a
few hundred yards) and for branches off a
network trunk, or backbone. RG-11 cable
is commonly used for this purpose.
M
Cable, Drop
Cable used to connect a network interface
card (NIC) to a transceiver on a thick Ether-
net network. Drop cable, also known as AUI
cable or transceiver cable, has a 15-pin AUI,
or DIX, connector at the NIC end and an
N-series connector at the transceiver end.
This term may also be applied loosely to
USING EXISTING COAXIAL CABLE
It may be tempting to try to use existing coaxial
cable-which is likely to be CATV cable-for
a network. If you're considering this, here's an
important point to keep in mind: Not all CATV
cables are the same.
Broadcast-oriented cables are designed to carry
video signals sent from one location in the net-
work, known as the head-end. Such cables are
designed for one-way communication, which
makes them useless for data networks. Even if
a bidirectional CATV cable is available, several
other considerations must be taken into account
before you can use this cable for a local-area
network.
If the cable will still be used to transmit TV chan-
nels, you need to find two frequency bands that
won't be used for TV channels. Each of these
bands must have at least 18 MHz band width. The
bands are used by a modem, which modulates
network data into the appropriate frequency
band at one end. A second modem demodulates
this signal at the other end. The TV and data net-
works will be independent of each other.
Because your network may be grafted onto an
existing CATV topology, you need to make sure
your system can deal with this. Typically, a CATV
network uses a tree topology. The head-end is
the root, and the signal is transmitted along suc-
cessive branches. For this setup, you need to
make sure that limitations on cable length are not
exceeded.
Tools for Working with Coaxial Cable


Prev Page 151 Next

Cable, Fiber-Optic
139
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
other cables that connect a network node
to a wiring center of some sort.
SEE ALSO
Cable
M
Cable, Feeder
A 25-pair cable that can be used for carrying
both voice and data signals. This cable can
run from equipment to distribution frame.
MCable, Fiber-Optic
Fiber-optic cable, also known as optical
fiber, provides a medium for signals using
light rather than electricity. Cables of this
type differ in their physical dimensions and
composition and in the wavelength(s) of
light with which the cable transmits. The fig-
ure "Context and properties of fiber-optic
cable" summarizes the features of this type
of cable.
Because fiber-optic communication uses
light signals, transmissions are not subject
to electromagnetic interference. This, and
the fact that a light signal encounters little
CONTEXT AND PROPER TIES OF FIBER-OPTIC CABLE
Context
Cable
Electrical

Twisted-Pair

Coaxial
Optical

Fiber-Optic
Fiber-Optic Properties
Medium for light signals
Light at certain wavelengths is best for signaling purposes
Comes in single-mode (thin fiber core; single light path) and multi-mode (thick fiber core; multiple light paths) versions
Multimode can be step-index or graded-index
Cable is very lightweight
Very high bandwidth
Immune to electromagnetic inteference, eavesdropping
Very long cable segments possible
FDDI networks
Long-haul lines
To connect network segments or networks
To connect mainframes to peripherals
To connect high-speed, high-performance workstations
Fiber-Optic Uses
Fiber-Optic Properties


Prev Page 152 Next

140
Cable, Fiber-Optic
resistance on its path (relative to an electri-
cal signal traveling along a copper wire),
means that fiber-optic cable can be used for
much longer distances before the signal must
be cleaned and boosted.
Some fiber-optic segments can be several
kilometers long before a repeater is needed.
In fact, scientists have sent signals over fiber-
optic lines for thousands of kilometers with-
out any signal boosters. In 1990, researchers
sent a 1 gigabit per second (Gbps) signal
almost 8,000 kilometers (about 5,000 miles)
without a boost!
In principle, data transmission using fiber
optics is many times faster than with electri-
cal methods. Speeds of over 10 Gbps are
possible with fiber-optic cable. In practice,
however, this advantage is still more promise
than reality, because the cable is waiting for
the transmission and reception technology
to catch up.
Nevertheless, fiber-optic connections
deliver more reliable transmissions over
greater distances, although at a some-
what greater cost. Fiber-optic cables cover
a considerable price and performance range. Fiber-Optic Core and Cladding
Currently, fiber-optic cable is used less often
to create a network than to connect two net-
works or network segments. For example,
cable that must run between floors is often
fiber-optic cable, most commonly of the
62.5/125 variety with an LED (light-
emitting diode) as the light source.
Being impervious to electromagnetic
interference, fiber is ideal for such uses
because the cable is often run through the
elevator shaft, and the elevator motor puts
Uses of Fiber-Optic Cable
out strong interference when the elevator is
running.
One reason fiber-optic networks are slow
to catch on is price. Network interface cards
(NICs) for fiber-optic nodes can cost several
thousand dollars, compared to street prices
of about $100 for some Ethernet and ARC-
net cards. However, when selecting optical
fiber, it is not always necessary to use the
most expensive fiber-optic connections. For
short distances and slower bandwidths,
inexpensive cable is just fine. In general, a
fiber-optic cable will always allow a longer
transmission than a copper cable segment.
The major components of a fiber-optic cable
are the core, cladding, buffer, strength mem-
bers, and jacket. Some types of fiber-optic
cable even include a conductive copper wire.
This can be used to provide power; for
example, to a repeater. The figure "Compo-
nents of a fiber-optic cable" illustrates the
makeup of this type of cable.
The core of fiber-optic cable consists of one
or more glass or plastic fibers through which
the light signal moves. Plastic is easier to
manufacture and use but works over shorter
distances than glass. The core can be any-
where from about 2 to several hundred
microns. (A micron, also known as a
micrometer, is a millionth of a meter, or
about 1/25,000 of an inch.)
In networking contexts, the most popular
core sizes are 60, 62.5, and 100 microns.
Most of the fiber-optic cable used in net-
working has two core fibers: one for com-
municating in each direction.
Fiber-Optic Cable Components


Prev Page 153 Next

Cable, Fiber-Optic
141
a
b
c
d
efghijklmnopqr
s
t
u
v
w
x
y
z
The core and cladding are actually manu-
factured as a single unit. The cladding is a
protective layer (usually of plastic) with
a lower index of refraction than the core.
The lower index means that light that hits
the core walls will be redirected back to
continue on its path. The cladding will be
anywhere between a hundred microns and a
millimeter (1000 microns) or so.
Fiber-Optic Buffer
The buffer of a fiber-optic cable is one
or more layers of plastic surrounding the
cladding. The buffer helps strengthen the
cable, thereby decreasing the likelihood of
micro-cracks, which can eventually grow
into larger breaks in the cable. The buffer
also protects the core and cladding from
potential corrosion by water or other mate-
rials in the operating environment. The
buffer can double the diameter of some
cable.
COMPONENTS OF A FIBER-OPTIC CABLE


Prev Page 154 Next

142
Cable, Fiber-Optic
A buffer can be loose or tight. A loose
buffer is a rigid tube of plastic with one or
more fibers (consisting of core and cladding)
running through it. The tube takes on all the
stresses applied to the cable, buffering the
fiber from these stresses. A tight buffer fits
snugly around the fiber(s). A tight buffer
can protect the fibers from stress due to
pressure and impact, but not from changes
in temperature.
Fiber-optic cable also has strength members,
which are strands of very tough material
(such as steel, fiberglass, or Kevlar) that pro-
vide extra strength for the cable. Each of the
substances has advantages and drawbacks.
For example, steel attracts lightning, which
will not disrupt an optical signal but may
seriously disrupt the people or machines
sending or receiving such a signal.
The jacket of a fiber-optic cable is an outer
casing that can be plenum or nonplenum, as
with electrical cable. In cable used for net-
working, the jacket usually houses at least
two fiber/cladding pairs: one for each
direction.
Fiber-optic cable can be either single-mode
or multimode. (Modes are the possible paths
for the light through a cable.)
Strength Members
Fiber-Optic Jacket
Single-Mode versus Multimode Cable
In single-mode fiber-optic cable, the core is
so narrow (generally less than 10 microns)
that the light can take only a single path
through it. Single-mode fiber has the least
signal attenuation, usually less than 2 deci-
bels (dB) per kilometer. This type of cable
is the most difficult to install, because it
requires the greatest precision, and it is the
most expensive of the major fiber-optic
types. However, transmission speeds of 50
Gbps and higher are possible. To get a sense
of this magnitude, note that a 10 Gbps line
can carry 130,000 voice channels.
Even though the core of single-mode
cable is shrunk to very small sizes, the clad-
ding is not reduced accordingly, nor should
it be. For single-mode fiber, the cladding
diameter should be about ten times the core
diameter. This ratio makes it possible to
make the cladding the same size as for popu-
lar multimode fiber-optic cable. This helps
create a de facto size standard. Keeping
the cladding large also makes the fiber and
cable easier to handle and more resistant to
damage.
Multimode fiber-optic cable has a wider
core, so that a beam of light has room to
follow multiple paths through the core.
Multiple modes (light paths) in a transmis-
sion produce signal distortion at the receiv-
ing end.
One measure of signal distortion is modal
dispersion, which is represented in nanosec-
onds (billionths of a second) of tail per kilo-
meter (ns/km). This value represents the
difference in arrival time between the fastest
Single-Mode Cable
Multimode Cable


Prev Page 155 Next

Cable, Fiber-Optic
143
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
and slowest of the alternate light paths. The Graded-Index Cable
value also imposes an upper limit on the
bandwidth, since the duration of a signal
must be larger than the nanoseconds of a tail
value. With step-index fiber, expect between
15 and 30 ns/km. Note that a modal disper-
sion of 20 ns/km yields a bandwidth of less
than 50 Mbps.
One reason optical fiber makes such a good
transmission medium is because the different
indexes of refraction for the cladding and
core help to contain the light signal within
the core. Cable can be constructed by chang-
ing abruptly from the core refractive index
to that of the cladding, or this change can be
made gradually. The two major types of
multimode fiber differ in this feature.
Cable with an abrupt change in refraction
index is called step-index cable. In step-
index cable, the change is made in a single
step. Single-step multimode cable uses this
method, and it is the simplest, least expen-
sive type of fiber-optic cable. It is also the
easiest to install. The core is usually between
50 and 125 microns in diameter; the clad-
ding is at least 140 microns.
The core width gives light quite a bit of
room to bounce around in, and the attenua-
tion is high (at least for fiber-optic cable):
between 10 and 50 dB/km. Transmission
speeds between 200 Mbps and 3 Gbps are
possible, but actual speeds are much lower.
Gradation of Refraction: Step-
Index Cable versus Graded-Index Cable
Step-Index Cable
Cable with a gradual change in refraction
index is called graded-index cable, or
graded-index multimode. This fiber-optic
cable type has a relatively wide core, like
single-step multimode cable. The change
occurs gradually and involves several layers,
each with a slightly lower index of refrac-
tion. A gradation of refraction indexes con-
trols the light signal better than the step-
index method. As a result, the attenuation is
lower, usually less than 15 dB/km. Similarly,
the modal dispersion can be 1 ns/km and
lower, which allows more than ten times the
bandwidth of step-index cable. Graded-
index multimode cable is the most com-
monly used type for network wiring.
Fiber core and cladding may be made of
plastic or glass. The following list summa-
rizes the composition combinations, going
from highest quality to lowest:
Single-mode glass: Has a narrow core,
so only one signal can travel through.
Graded-index glass: Not tight enough to
be single-mode, but the gradual change
in refractive index helps give more
control over the light signal.
Step-index glass: The abrupt change from
the refractive index of the core to that
of the cladding means the signal is less
controllable.
Plastic-coated silica (PCS): Has a rela-
tively wide core (200 microns) and a
relatively low bandwidth (20 MHz).
Fiber Composition


Prev Page 156 Next

144
Cable, Fiber-Optic
Plastic: This should be used only for very
short distances.
To summarize, fiber-optic cables may
consist of glass core and glass cladding (the
best available). Glass yields much higher
performance, in the form of higher band-
width over greater distances. Single-mode
glass with a small core is the highest quality.
Cables may also consist of glass core and
plastic cladding. Finally, the lowest grade
fiber composition is plastic core and plastic
cladding. Step-index plastic is at the bottom
of the heap in performance.
Fiber-optic cables are specified in terms
of their core and cladding diameters. For
example, a 62.5/125 cable has a core with a
62.5 micron diameter and cladding with
twice that diameter.
The following are some commonly used
fiber-optic cable configurations:
FIBER-OPTIC CABLE QUALITY
Here are a few points about fiber-optic cable
(other things being equal):
I
The smaller the core, the better the signal.
I
Fiber made of glass is better than fiber made
of plastic.
I
The purer and cleaner the light, the better the
signal. (Pure, clean light is a single color, with
minimal spread around the color's primary
wavelength.)
I
Certain wavelengths of light behave better
than others.
Fiber-Optic Cable Designations
8/125: A single-mode cable with an 8
micron core and a 125 micron clad-
ding. This type of cable is expensive
and currently used only in contexts
where extremely large bandwidths are
needed (such as in some real-time
applications) or where large distances
are involved. An 8/125 cable configu-
ration is likely to broadcast at a light
wavelength of 1,300 or 1,550 nm.
62.5/125: The most popular fiber-
optic cable configuration, used in most
network applications. Both 850 and
1,300 nm wavelengths can be used
with this type of cable.
100/140: The configuration that IBM
first specified for fiber-optic wiring in
a Token Ring network. Because of the
tremendous popularity of the 62.5/125
configuration, IBM now supports both
configurations.
Make sure you buy fiber-optic cable
with the correct core size. If you know what
kind of network you plan to build, you may
be constrained to a particular core size.
IBM usually specifies a core of 100 microns
for Token Ring networks; other networks
more commonly use cable with a 62.5
micron core.
In addition to the cable itself, a fiber-optic
connection needs a light source to generate
the signal, as well as connectors, repeaters,
and couplers to route and deliver the signal.
The figure "Components of a fiber-optic
connection" illustrates how this works.
Components of a Fiber-Optic Connection
Transmitter


Prev Page 157 Next

Cable, Fiber-Optic
145
a
b
c
d
efghijklmno
p
q
r
s
t
u
v
w
x
y
z
Fiber-optic transmitters convert an elec-
tronic signal into light and send this light
signal into the fiber core. The transmitter's
light source and output optical power are
crucial elements in determining the transmit-
ter's performance.
The transmitter's output power depends
on several things, including the fiber and
cladding sizes and the fiber's numerical aper-
ture (NA). The NA is a measure of the fiber's
ability to gather light and is determined by
the angle over which light hitting the fiber
will move through it.
Output power values range from less than
50 to over 200 microwatts. Smaller cores
generally have lower output power, but also
less signal attenuation and higher band-
width. Output power values should not be
too high, since this increases energy require-
ments and also risks frying the components
at the receiving end.
Transmitters use either digital or
analog modulation. Analog modulation
is used for voice, video, and even radar
signals, which require bandwidths ranging
from tens of kilohertz to hundreds of mega-
hertz, and even as high as a gigahertz.
Digital modulation is used in computer net-
works and in long-haul telephone systems,
COMPONENTS OF A FIBER-OPTIC CONNECTION


Prev Page 158 Next

146
Cable, Fiber-Optic
which require transmission speeds ranging
from tens of kilobits to more than a gigabit
per second. Transmitters differ in speed. Not
surprisingly, the faster ones are also more
expensive.
The light source will be a laser or a light-
emitting diode (LED). A good light source
in a fiber-optic connection should have the
following characteristics:
I Fast rise and fall times. The rise time is
the time required for a light source to
go from 10 to 90 percent of the desired
level. This time limits the maximum
transmission rate, so it should be as
short as possible. Lasers have a rise
time of less than a nanosecond; the
rise time for LEDs ranges from a few
nanoseconds to a few hundred
nanoseconds.
I A narrow spectral width. The spectral
width refers to the range of wave-
lengths emitted by the light source, and
it should be as narrow as possible.
Spectral widths for lasers are 1 to 3
nm; for LEDs, they are from 30 to
50 nm.
I Light emission at a central wavelength
with minimal spectral width. The cen-
tral wavelength is the primary wave-
length of the light being emitted. For
various reasons, wavelengths of 820,
850, 1300, and 1550 nm have all been
used. LEDs are used for the first three
of the wavelengths, but rarely for 1550
nm. Lasers can be used at all of these
wavelengths, and single-frequency
lasers (possible at the two highest
wavelengths) make it possible to emit
at a particular wavelength with mini-
mal spectral width.
I A good relationship between the emit-
ting area and acceptance angle. The
emitting area is the opening through
which the transmitter emits its light.
This should be small in relation to the
fiber core's acceptance angle, so that
all the light emitted by the transmitter
will enter the core. Not surprisingly,
lasers have a much smaller emitting
area than LEDs.
I Steady, strong output power. The
higher the output power, the stronger
the signal and the further it can travel
without becoming too weak. Laser
output can be as much as 1000 times
that of LEDs.
I A long lifetime. The lifetime of a light
source is the amount of time before the
source's peak output power is half its
original level. This is generally in the
millions of hours (longer than ours)
and is typically longer for LEDs than
for lasers!
Although lasers are clearly the light
source of choice, LEDs are generally the
light source of record. The most likely rea-
son for this is price; transmitters that use
LEDs are usually much less expensive. This
is not a problem for networking purposes,
however, because LEDs operating at 820 or
850 nm are fine for the short-distance, fiber-
optic connections currently most popular.
Despite their performance shortcomings
Light Source


Prev Page 159 Next

Cable, Fiber-Optic
147
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
compared with lasers, LEDs are more
reliable and less prone to breakdowns.
Fiber-optic receivers undo the work of trans-
mitters: they accept a light signal and con-
vert this to an electrical signal representing
information in analog or digital form. A
receiver's performance depends on how well
its three main components work. The fol-
lowing are the main components of a fiber-
optic receiver:
I The photodetector, which "sees" the
optical signal and converts it into elec-
trical form. This produces a current
that is proportional to the level of light
detected.
I The amplifier, which boosts the sig-
nal and gets it into a form ready for
processing.
I
The processor, which tries to repro-
duce the original signal.
The receiver also includes interfaces for
the cable carrying the light signal and the
device to which the electrical signal is being
passed.
The photodetector and amplifier pro-
cesses are essentially identical for analog and
digital signals. The main differences are in
the processor.
There are several classes of photodetec-
tors, each suitable for different speed and
distance configurations. The receiver sensi-
tivity specifies the weakest signal that the
photodetector can detect. This information
may be expressed as an absolute value, such
as 10 microwatts, or as a microwatt level
needed for a given bit error rate (BER).
Duty Cycle A duty cycle specifies the ratio
of high to low signal values in a digital
transmission. This is not necessarily equal to
the proportion of 0 and 1 bit values in the
message, because some signal-encoding
methods will encode a 1 as high at one point
in a transmission and as low in another
point. (See the Encoding, Signal article for
examples of such methods.) The ideal duty
cycle is 50 percent.
The duty-cycle value is important because
receivers use a reference level as the thresh-
old between high and low values. Some
receivers adjust this reference during a trans-
mission. If a duty-cycle value deviates from
the 50 percent ideal, the altered threshold
level could lead to more erroneous values.
For example, if a threshold is adjusted
downward because of a 20 percent duty
cycle, low signals that are marginally but not
significantly higher than normal may be mis-
interpreted as high values. There are two
strategies for getting around the potential
error problem: signal encoding and reference
levels.
Certain signal-encoding methods, such as
the Manchester and differential Manchester
methods used in Ethernet and Token Ring
networks, always have a 50 percent duty
cycle. The tradeoff for this nice behavior
is that these encoding methods require a
clock that runs at twice the data rate (since
every interval is associated with two electri-
cal levels).
It is possible to build a receiver that has
an absolute reference level; that is, one that
will always correspond to the level of a
Receiver


Prev Page 160 Next

148
Cable, Fiber-Optic
50 percent duty cycle. This is accomplished
by coupling the receiver to a DC power sup-
ply. The tradeoff for this is that the receiver
has higher power requirements; it requires a
signal that is 6 to 8 dB (roughly, four to
eight times) stronger than for an ordinary
receiver.
A fiber-optic transceiver includes both a
transmitter and a receiver in the same com-
ponent. These are arranged in parallel so
that they can operate independently of each
other. Both the receiver and the transmitter
have their own circuitry, so that the com-
ponent can handle transmissions in both
directions.
Like a transceiver, a fiber-optic repeater
includes both a transmitter and a receiver
in the same component. However, in the
repeater, these components are arranged in
series, separated by circuitry for cleaning
and boosting the signal. The receiver gets the
signal and passes it through the booster to
the transmitter.
Connectors serve to link two segments of
cable or a cable and a device. A connector is
used for temporary links. To link two sec-
tions of cable permanently, use a splice; to
link more than two sections of cable, use a
coupler. In general, use a splice when possi-
ble; use a connector when necessary.
A good connector or splice should have
the following properties:
I Low power loss. There should be mini-
mal loss of signal power going across
the connection or splice. For networks
and short-distance connections, the
loss should be less than 1 dB; for long-
haul connections, there should be less
than 0.2 dB loss.
I Durability. The connector should be
capable of multiple matings (connec-
tions) without loosening or becoming
unreliable. Durability values typically
range between about 250 and 1000
matings.
I Ease of use. The connector or splice
should be easy to install.
I Low price. The less expensive, the bet-
ter, provided all the preceding features
are satisfactory.
There are many types of connector
designs used for fiber-optic cable. Some of
the most commonly used ones in networking
are SC, ST, SMA, and the MIC connector
specified for the FDDI (Fiber Distributed
Data Interface) network architecture. See the
Connectors, Fiber-Optic article for more
information about fiber-optic connectors.
If a fiber-optic connection is more or less
permanent, it may make more sense to splice
the cable segments together. Splicing tech-
niques are more reliable and precise than
connectors. Because of this, signal loss at
splices is much lower (almost always less
than 1 dB, and often less than 0.25 dB) than
at connectors. Splicing is almost always used
for long-haul, fiber-optic cable.
Fusion and mechanical splices are the two
most common splicing methods. Of the two,
fusion gives the better splices.
Transceiver
Repeater
Connectors and Splices


Prev Page 161 Next

Cable, Fiber-Optic
149
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
A fusion splice welds the two fibers
together using a high-precision instrument.
This type of splice produces losses smaller
than 0.1 dB. The equipment for such splic-
ing is quite expensive, however.
A mechanical splice is accomplished by
fitting a special device over the two fibers to
connect them and lock them into place. The
device remains attached to the splice area
to protect the splice from environmental
effects, such as moisture or pressure.
Mechanical splices have higher signal losses
than fusion splices, but these losses may still
be less than 0.25 dB.
Fiber-optic couplers route an incoming sig-
nal to two or more outgoing paths. Coup-
lers are needed in fiber-optic networks.
When an electrical signal is split and sent
along parallel paths, each derived signal is
the same strength. This is not the case with
light signals.
After the signal is split, the derived opti-
cal signals are each weaker than the original
signal. For example, if a fiber-optic coupler
splits a signal into two equal signals, each of
those derived signals loses 3 dB relative to
the original signal, just from the signal halv-
ing. Couplers can be designed to split a sig-
nal equally or unequally. See the Coupler,
Fiber-Optic article for more information.
Couplers used in networks need some type
of bypass mechanism, so that the coupler
can be disconnected if the coupler's target
nodes are not on the network. This discon-
nection capability is accomplished with
an optical switch, which allows the light
to bypass a node and continue around the
network.
As mentioned earlier, light signals can be
diminished by coupling. In addition, factors
that contribute to signal loss across a stretch
of cable include the following:
Pulse dispersion: If the cable's core width
is large compared with the light's
wavelength, light enters the core at dif-
ferent angles and will travel different
distances to the destination. As
explained earlier, the difference in
arrival times between the fastest and
slowest signals in a group is measured
in nanoseconds of tail over the dis-
tance the light must travel. This value
limits the maximum transmission rate,
because signal pulses must be sepa-
rated by at least the nanoseconds of
tail time. For example, if a signal
acquires 10 nanoseconds of tail over
the required distance, the maximum
transmission rate is 100 Mbps.
Attenuation: Loss of signal strength that
occurs because some of the light is
absorbed by the cladding, and some
light is scattered as a result of imper-
fections in the fiber.
Fiber bending: Signal loss can occur
because the fiber is bent in particular
ways. Multiple bands of light (known
as modes) enter a core, each at slightly
different angles. Bending the fiber can
enable certain modes to escape from
the core. Since the modes that escape
Couplers
Optical Switches
Fiber-Optic Cable Signal Loss


Prev Page 162 Next

150
Cable, Fiber-Optic
will not be random, fiber bending can
introduce systematic loss of certain sig-
nal components. Simply rolling fiber
cable onto a spool for distribution can
introduce fiber bending. Cable manu-
facturers design their cable spools
carefully, and some even publish speci-
fications for the spool.
Microbending: Microbends are tiny kinks
that can arise in the cable as a result of
various stresses (for example, attach-
ing a connector at the end of a cable).
Microbends in the fiber can cumulate,
and the presence of many kinks can
significantly increase the signal loss
from bending.
Fiber ovality: If the fiber's core and clad-
ding are not round, the nonuniform
shape will distort the signal. This can
happen, for example, if the cable was
squashed with a heavy weight, so that
the core and cladding are partially
flattened.
Fiber-optic connections offer the following
advantages over other types of cabling
systems:
I Light signals are impervious to
interference from EMI or electrical
crosstalk. Light signals do not interfere
with other signals. As a result, fiber-
optic connections can be used in
extremely adverse environments, such
as in elevator shafts or assembly
plants, where powerful motors and
engines produce lots of electrical noise.
Advantages of Fiber-Optic Cable
I Fiber-optic lines are much harder to
tap, so they are more secure for private
lines.
I
Light has a much higher bandwidth, or
maximum data-transfer rate, than elec-
trical connections. (This speed advan-
tage has yet to be realized in practice,
however.)
I
The signal has a much lower loss rate,
so it can be transmitted much further
than it could be with coaxial or
twisted-pair cable before boosting is
necessary.
I Optical fiber is much safer, because
there is no electricity and so no danger
of electrical shock or other electrical
accidents.
I Fiber-optic cable is generally much
thinner and lighter than electrical
cable, and so it can be installed more
unobtrusively. (Fiber-optic cable
weighs about an ounce per meter;
coaxial cable weighs nearly ten times
that much.)
I Cable making and installation are
much easier than they were in the early
days.
The disadvantages of fiber-optic connections
include the following:
I
Fiber-optic cable is currently more
expensive than other types of cable.
I Other components, particularly NICs,
are very expensive.
Disadvantages of Fiber-Optic Cable


Prev Page 163 Next

Cable, Fiber-Optic
151
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
I Certain components, particularly cou-
plers, are subject to optical crosstalk.
I Fiber connectors are not designed to
be used as often as you would like.
Generally, they are designed for fewer
than a thousand matings. After that,
the connection may become loose,
unstable, or misaligned. The resulting
signal loss may be unacceptably high.
I Many more parts can break in a
fiber-optic connection than in an
electrical one.
It is only fitting that the most complex
wiring technology should also have the
most sophisticated tools. Optical fiber
undergoes an extensive set of tests and
quality-control inspections before it
even leaves the manufacturer.
The manufacturers' tests are designed to
get complete details about the cable's physi-
cal and optical properties. Optical properties
include attenuation, dispersion, and refrac-
tive indexes of the core and cladding layers.
Physical properties include core and clad-
ding dimensions, numerical aperture and
emitting areas, tensile strength, and changes
in performance under extreme temperature
and/or humidity conditions (or as a result of
repeated changes in temperature). The val-
ues for these properties are used to evaluate
cable performance.
The equipment you might need to test
fiber-optic cables in a network setting
includes the following:
I An installation kit-a general-purpose
tool set for dealing with optical fiber.
Fiber-Optic Cable Tools
Such a toolkit will include cable strip-
pers, scissors, crimping tools, epoxy,
pliers, canned air (for cleaning fibers
after polishing), inspection micro-
scope, polishing materials, and so on.
I Optical power meter, which is a device
that can read levels of optical signals
on a fiber-optic line. Using sensors
attached to the cable, this device can
report absolute or relative signal levels
over a range of 110 dB (which means
that the weakest and strongest detect-
able signals differ by a factor of over
10 billion). An optical power meter
can also be used to measure light at
specific wavelengths.
I An OTDR (optical time domain reflec-
tometer), which is a device that can
measure the behavior of the light sig-
nals over time and create graphical
representations of these measurements.
An OTDR can be used to measure
signal loss along a stretch of cable and
to help locate a fault in a fiber-optic
connection.
I Splicer, which is used to create splices,
or permanent connections in an optical
fiber. Fusion splicers are the most
expensive devices of this sort.
I Polishers, which are used to prepare
fiber ends for splicing or connection.
I A microscope, so you can inspect the
results of a splicing or polishing opera-
tion. A microscope may be included in
an installation toolkit.


Prev Page 164 Next

152
Cable, Horizontal
Many vendors sell both electrical and
fiber-optic cable, as well as connectors,
installation, and testing tools. The
following vendors offer an extensive
selection of fiber-optics products. (See
the Cable article for other cable vendors.)
AMP Incorporated: (800) 522-6752;
(717) 564-0100
CSP (Computer System Products): (800)
422-2537; (612) 476-6866; Fax (612)
476-6966
FIS (Fiber Instrument Sales): (800) 445-
2901; (315) 736-2206; Fax (315)
736-2285
SEE ALSO
Cable; Cable, Coaxial; Cable, Twisted-
Pair; Connector, Fiber-Optic; Coupler,
Fiber-Optic; FDDI (Fiber Distributed
Data Interface)
Fiber-Optic Cable Vendors
MCable, Horizontal
Horizontal cable is defined by the EIA/TIA-
568 committee as any cable that goes from a
wiring closet, or distribution frame, to the
wall outlet in the work area. Distribution
frames from a floor or building are con-
nected to other frames using backbone
cable.
In a sense, horizontal cable is the most
crucial in the entire network cabling struc-
ture. Since it is installed in the walls, floors,
ceiling, or ground, the installation process
can be difficult and expensive. Moreover, the
cable should be able to handle future stan-
dards and technology.
The EIA/TIA-568 recognizes four main
types of horizontal cable, and several
optional variants. These types are listed in
the table "EIA/TIA-568 Main and Optional
Types of Horizontal Cable." The EIA/TIA
specifications call for at least two cables
EIA/TIA-568 MAIN AND OPTIONAL
TYPES OF HORIZONTAL CABLE
CABLE TYPE
MAIN
OPTIONAL
UTP
STP
Coaxial
Optical fiber
Undercarpet
100-ohm, four-pair UTP cable
150-ohm STP cable, such as that defined in
the IBM Cable System (ICS)
50-ohm, thin coaxial cable, such as the cable 75-ohm (broadband) coaxial cable, such as
used in thin Ethernet networks
62.5/125-micron (step- or graded-index)
multimode optical fiber
100-ohm, 25-wire-pair UTP cable
100-ohm STP cable
CATV cable
Multimode fiber with other core/cladding
ratios of 50/125-micron, 100/140-micron,
etc.
Flat cable (such as Type 8 in the ICS) that
can be run under carpet without posing a
hazard


Prev Page 165 Next

Cable, IBM
153
a
b
c
d
ef
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
from this list to be run to every wall outlet.
At least one of these should be unshielded
twisted-pair (UTP).
COMPARE
Cable, Backbone
SEE ALSO
Cable
MCable, IBM
The IBM Cable System (ICS) was designed
by IBM for use in its Token Ring networks
and also for general-purpose premises wir-
ing. The figure "Context and properties of
the IBM Cable System" summarizes the fea-
tures of this type of cable.
CONTEXT AND PROPER TIES OF THE IBM CABLE SYSTEM
Context
Cable
Electrical

Twisted-Pair

Coaxial
Optical

Fiber-Optic


Comprises Types 1 through 9 (of which all types but 4 and 7 are defined)
Type 5 is fiber-optic
Type 3 is unshielded twisted-pair (UTP)
Remaining types are shielded twisted-pair (STP)
Type 1 is most common in Token Ring Networks
Type 3 is not recommended for 16 Mbps networks
Type 3 cable generally requires a media filter
Type 6 is used mainly as short-distance patch cable
Type 8 is flat cable for use under a carpet
IBM Token Ring networks
10BaseT Ethernet networks
ARCnet networks
ISDN lines
Some IBM 3270 networks
IBM Cable System Uses
IBM Cable System Properties



Prev Page 166 Next

154
Cable, IBM
IBM has specified nine types of cable,
mainly twisted-pair, but with more stringent
specifications than for the generic twisted-
pair cabling. The type taxonomy also
includes fiber-optic cable, but excludes co-
axial cable. The twisted-pair versions differ
in the following ways:
I Whether the type is shielded or
unshielded
I Whether the carrier wire is solid
or stranded
I The gauge (diameter) of the carrier
wire
I The number of twisted pairs
Specifications have been created for seven
of the nine types. Types 4 and 7 are unde-
fined; presumably, they are reserved for
future use.
Type 1 cable is shielded twisted-pair (STP),
with two pairs of 22-gauge solid wire. It is
used for data-quality transmission in IBM's
Token Ring network. It can be used for the
main ring or to connect lobes (nodes) to
multistation attachment units (MAUs),
which are wiring centers.
Although not required by the specifica-
tions, a plenum version is also available, at
about twice the cost of the nonplenum cable.
Compare Type 1 with Type 6.
Type 2 is a hybrid consisting of four pairs of
unshielded 22-gauge solid wire (for voice
Type 1 Cable
Type 2 Cable
transmission) and two pairs of shielded 22-
gauge solid wire (for data). Although not
required by the specifications, a plenum
version is also available, at about twice
the cost.
Type 3 is unshielded twisted-pair (UTP),
with two, three, or four pairs of 22- or 24-
gauge solid wire. The pairs have at least two
twists per foot. This category requires only
voice-grade capabilities, and so may be used
as telephone wire for voice transmissions.
Type 3 is not recommended for 16 Mbps
Token Ring networks.
Although not required by the specifica-
tions, a plenum version is also available, at
about twice the cost.
Type 3 cable is becoming more popular as
adapter cable, which is used to connect a
node to a MAU. You must use a media filter
if you are using Type 3 cable to connect a
node to a MAU or if you need to switch
between UTP and STP in a Token Ring net-
work. However, you should not mix Type 1
and 3 cable in the same ring. Mixing cable
types makes trouble-shooting difficult.
Some manufacturers offer higher-quality
Type 3 cable for greater reliability. Such
cable has more twists per foot, for greater
protection against interference. Many ven-
dors recommend that you use Category 4
cable (with 12 twists per foot). This category
of cable costs about 20 percent more than
ordinary Type 3 cable, but is rated for higher
speeds. The category value represents a clas-
sification system for the performance of UTP
cable. See the Cable Standards article for
more information.
Type 3 Cable


Prev Page 167 Next

Cable, Quadrax
155
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
Type 5 is fiber-optic cable, with two glass
fiber cores, each with a 100-micron diameter
and a 140-micron cladding diameter. (IBM
also allows the more widely used 62.5/125-
micron fiber.)
This type is used for the main ring
path (the main network cabling) in a Token
Ring network to connect MAUs over greater
distances or to connect network segments
between buildings. Plenum versions of
Type 5 cable are available at only a slightly
higher cost.
Type 6 is STP cable, with two pairs of 26-
gauge stranded wire. This type is commonly
used as an adapter cable to connect a node
to a MAU. In that type of connection, the
PC end of the cable has a male DB-9 or
DB-25 connector, and the MAU end has
a specially designed IBM data connector.
Type 6 cable is also used as a patch cable;
for example, to connect MAUs. For this
use, the cable has IBM data connectors at
each end.
Because Type 6 is used mostly for shorter
distances, the price per foot tends to be
higher than for other cable types.
Type 8 is STP cable, with two pairs of flat,
26-gauge solid wire. This type is specially
designed to be run under a carpet, so the
wires are flattened. This makes the cable
much more prone to signal loss than Type 1
or Type 2 cable; however, the performance
of Type 8 cable is adequate for the short
Type 5 Cable
Type 6 Cable
Type 8 Cable
distances usually involved in under-the-
carpet cabling.
Type 9 is STP cable, with two pairs of
26-gauge solid or stranded wire. This type
is covered with a plenum jacket and is
designed to be run between floors.
SEE ALSO
Cable, Twisted-Pair
MCable, Patch
Cable used to connect two hubs or multi-
station attachment units (MAUs). IBM Type
1 or Type 6 patch cables can be used for
Token Ring networks.
SEE ALSO
Cable, IBM
MCable, Plenum
Cable that has a fire-resistant jacket, which
will not burn, smoke, or give off toxic fumes
when exposed to heat. The cable goes
through a plenum, a conduit, or shaft, run-
ning inside a wall, floor, or ceiling. Fire regu-
lations generally stipulate that cable running
through such conduits must be fireproof.
SEE ALSO
Cable
M
Cable, Quadrax
A type of coaxial cable. Quadrax cable,
sometimes known simply as quadrax, is
a hybrid of triaxial and twinaxial cable. Like
twinaxial cable, quadrax has the extra
Type 9 Cable


Prev Page 168 Next

156
Cable, Quad Shield
carrier wire with dielectric; like triaxial
cable, quadrax has extra shielding.
SEE ALSO
Cable, Coaxial
M
Cable, Quad Shield
A type of coaxial cable with four layers of
shielding: alternating layers of foil and braid
shields. Quad shield cable, sometimes
known simply as quad shield, is used in
situations where heavy electrical interference
can occur, such as in industrial settings.
SEE ALSO
Cable, Coaxial
M
Cable, Riser
Cable that runs vertically; for example,
between floors in a building. Riser cable
often runs through available shafts (such as
for the elevator). In some cases, such areas
can be a source of electrical interference.
Consequently, optical fiber (which is imper-
vious to electromagnetic interference) is gen-
erally used as rise cable.
M
Cable Standards
Several cable standards are concerned with
the performance and reliability of cables
under actual working conditions. In particu-
lar, these standards specify the cable's mini-
mal acceptable behavior under adverse
working conditions; for example, in manu-
facturing or industrial environments, where
heavy machinery is turned on and off during
the course of operations. Such actions can
generate strong interference and power-
supply variations. Cable environments are
often distinguished in terms of the demands
made on the cable. The standards also spec-
ify the minimum behavior required under
extreme conditions, such as fire.
The most commonly used safety stan-
dards in the United States are those specified
in the National Electric Code and in docu-
ments from Underwriters Laboratories.
Other standards are specified by the Elec-
tronic Industries Association/Telecommuni-
cations Industries Association, Electrical
Testing Laboratory, and Manufacturing
Automation Protocol.
The NEC is published by the National Fire
Protection Agency (NFPA, 617-770-3000),
and specifies safety standards for general-
purpose cables in commercial and residential
environments, and also specifically for
cables used for communications. The
Class 2 (CL2x) standards apply to general-
purpose cables, and the Communications
(CMx) standards apply to special-purpose
cables capable of carrying data.
Of the CL2 standards, the most strin-
gent ones apply to Class 2 plenum cable
(CL2P). Cable that meets or exceeds these
standards is said to be CL2P compliant.
CMP-compliant cable meets the correspond-
ing standard for plenum communications
cable.
The less stringent CL2R standards apply
to riser cable (cable that can be used, for
example, in a vertical utility shaft between
floors in a building). The corresponding
standard for communications riser cable
is CMR.
Be wary if you intend to use cable that is
neither CMx- nor CL2x-compliant. Older
The National Electric Code (NEC)


Prev Page 169 Next

Cable Standards
157
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
cable that is already in the walls may be
noncompliant.
UL tests cable and other electrical devices to
determine the conditions under which the
cable or device will function safely and as
specified. UL-listed products have passed
safety tests performed by inspectors at the
Underwriters Laboratories.
Two tests are most directly relevant to
network cable:
UL-910: Tests smoke emissions and the
spread of flames for plenum cable.
This test corresponds to the CL2P level Association/Telecommunications
of safety standards. A cable that passes
the UL-910 test is rated as OFNP
(optical fiber, nonconductive plenum)
by UL.
UL-1666: Tests the performance of riser
cable in a fire. This test corresponds
roughly to the CL2R level of safety
standards. A cable that passes the UL-
1666 test is rated as OFNR (optical
fiber, nonconductive riser) by UL.
UL also uses a system of markings to cat-
egorize cable as falling into one of five levels
(I through V). Cables that meet level I and II
standards meet minimum UL safety require-
ments, but the performance of these cables
may be inadequate for networking purposes.
Cables that meet level III, IV, or V standards
meet both safety and various performance
requirements. Higher levels allow for less
attenuation and interference due to crosstalk
than lower levels.
Cable should be UL-listed, and just about
every cable is. However, you need to find out
which listing applies. For example, OFNR
Underwriters Laboratories (UL)
cable is UL-listed but is not suitable for envi-
ronments that demand fire protection.
For most networking applications, cable
that meets requirements for UL level III or
above should be adequate.
A committee for EIA/TIA has created yet
another classification system for specifying
the performance of unshielded twisted-pair
(UTP) cable. The EIA/TIA taxonomy
includes the following categories (1 through
5) whose criteria correspond roughly to
the performance criteria specified for the UL
levels:
Category 1: Voice-grade, UTP telephone
cable. This describes the cable that has
been used for years in telephone com-
munications. Officially, such cable is
not considered suitable for data-grade
transmissions (in which every bit must
get across correctly). In practice,
however, it works fine over short dis-
tances and under ordinary working
conditions.
Category 2: Data-grade UTP, capable of
supporting transmission rates of up to
4 megabits per second (Mbps). IBM
Type 3 cable falls into this category.
UNDERWRITERS LABORATORIES
(UL) PHONE NUMBERS
East Coast: (516) 271-6200
Central: (708) 272-8800
West Coast: (408) 985-2400
Electronic Industries
Industries Association (EIA/TIA)


Prev Page 170 Next

158
Cable Standards
Category 3: Data-grade UTP, capable of
supporting transmission rates of up to
10 Mbps. A 10BaseT network requires
such cable.
Category 4: Data-grade UTP, capable of
supporting transmission rates of up to
16 Mbps. A 16 Mbps IBM Token Ring
network requires such cable.
Category 5: Data-grade UTP, capable of
supporting transmission rates of up to
155 Mbps (but officially only up to
100 Mbps). The proposed CDDI
(Copper Distributed Data Interface)
networks and 100Base-X network
architecture require such cable.
Many cable vendors also use a five-level sys-
tem to categorize their UTP cable. Just as
there is overlap in the paths to enlighten-
ment in various religious traditions, there is
some overlap between these levels and the
other systems discussed here. For example,
the references to Level 4, Category 4 cable
identify the cable according to the features
described here and also according to the
features in the EIA/TIA specifications.
Level 1: Voice-grade cable, which is suit-
able for use in the "plain old telephone Electrical Testing Laboratory (ETL)
system" (or POTS). Such cable can
handle data at up to 1 Mbps.
Level 2: Data-grade cable that is capable
of transmission speeds as high as 4
Mbps. This level corresponds roughly
to the Type 3 cable described in IBM's Manufacturing Automation Protocol (MAP)
Cabling System (see the Cable, IBM
article). Level 2 cable also meets
the requirements for the 1Base5
(StarLAN) Ethernet network devel-
oped by AT&T.
Level 3: Data-grade cable that is capable
of transmission speeds as high as
16 Mbps. This level corresponds to
Category 3 cable in the EIA/TIA-568
specifications. Level 3 cable is used in
4 Mbps or 16 Mbps Token Ring net-
works, and also in 10BaseT Ethernet/
802.3 networks.
Level 4: Data-grade cable that is capable
of transmission speeds as high as 20
Mbps. This level corresponds to Cate-
gory 4 cable in the EIA/TIA-568 speci-
fications. Level 4 cable is used for
ARCnet Plus, a 20 Mbps version of the
ARCnet network architecture.
Level 5: Data-grade cable that is capable
of transmission speeds as high as 100
Mbps. This level corresponds to Cate-
gory 5 cable in the EIA/TIA-568 speci-
fications. Level 5 cable is used for
CDDI (or TPDDI), which are copper-
based implementations of the 100
Mbps FDDI network architecture.
100Base/X, a proposed 100 Mbps ver-
sion of Ethernet, is also intended to
run on this type of cable.
The ETL is an independent laboratory that
tests and rates products for manufacturers.
Vendors specify if their cable has been tested
and verified by ETL.
The most commonly observed performance
standards, arguably, are those associated
Performance Levels


Prev Page 171 Next

Cable, Twisted-Pair
159
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
with the MAP. Among other things, this
standard specifies the expected performance
for cables in the highly automated and
machinery-heavy industrial working envi-
ronments of the future.
Cable that meets MAP standards gener-
ally has quad shields; that is, four layers of
shielding around the central core in a coax-
ial cable. The four layers of shielding help
protect the cable against signal loss from the
conductor wire and against electromagnetic
interference from the outside world; for
example, from heavy machinery being
turned on and off. See the MAP article
for more information.
M
Cable Tester
An instrument for testing the integrity and
performance of a stretch of cable. Cable
testers run various tests to determine the
cable's attenuation, resistance, characteristic
impedance, and so on. High-end testers can
test cable for conformity to various network MCable, Twisted-Pair
architecture specifications, and can some-
times even identify a particular type of cable.
M
Cable, Transceiver
Cable used to connect a network interface
card to a transceiver, mainly in Ethernet
architectures. A transceiver cable usually
has an AUI connector at one end and an
N-series or other type of connector at the
other end. Coaxial transceiver cable comes
in thick and thin versions. You can also get
special cable with a built-in right angle.
MCable, Triaxial
A type of coaxial cable. Also called triax,
this cable has an inner braid surrounded by
an inner (nonplenum) jacket, surrounded
by an outer copper braid. The extra shield-
ing makes a big difference because of the
grounding and improved protection.
SEE ALSO
Cable, Coaxial
M
Cable, Twinaxial
A type of coaxial cable. Also called twinax,
this cable has two insulated carrier wires,
generally twisted around each other, which
helps cut down considerably on magnetic
interference. Twinaxial cables are used in
IBM and AppleTalk networks.
SEE ALSO
Cable, Coaxial
Twisted-pair cable is very widely used, inex-
pensive, and easy to install. It can transmit
data at an acceptable rate (up to 100 Mbps
in some network architectures). The best-
known example of twisted-pair wiring is
probably telephone cable, which is
unshielded and is usually voice-grade, rather
than the higher-quality data-grade cable
used for networks. The figure "Context and
properties of twisted-pair cable" summa-
rizes the features of this type of cable.
In a twisted-pair cable, two conductor
wires are wrapped around each other. A sig-
nal is transmitted as a differential between
the two conductor wires. This type of signal


Prev Page 172 Next

160
Cable, Twisted-Pair
CONTEXT AND PROPER TIES OF TWISTED-PAIR CABLE
Context
Cable
Electrical

Twisted-Pair

Coaxial
Optical

Fiber-Optic

Includes shield around twisted pairs
150 ohm impedance
Information in differential signal between wires in a pair
Subject to near-end crosstalk (NEXT)
Subject to electromagnetic interference
Generally uses RJ-xx connectors
IBM Token Ring networks
ARCnet networks
Rarely in Ethernet networks
Shielded Twisted-Pair (STP) Properties
Shielded Twisted-Pair (STP) Uses
No shield around twisted pairs
100 ohm impedance
Information in differential signal between wires in a pair
Subject to near-end crosstalk (NEXT)
Subject to electromagnetic interference
Generally uses RJ-xx connectors
Performance grades specified in EIA/TIA-568 CATEGORIES I-5
Unshielded Twisted-Pair (UTP) Properties
10BaseT Ethernet networks
ARCnet networks
Certain sections of IBM Token Ring networks
Telephone lines (voice-grade)
Unshielded Twisted-Pair (UTP) Uses


Prev Page 173 Next

Cable, Twisted-Pair
161
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
is less prone to interference and attenuation,
because using a differential essentially gives
a double signal, but cancels out the random
interference on each wire.
Twisting within a pair minimizes cross-
talk between pairs. The twists also help deal
with electromagnetic interference (EMI) and
radio frequency interference (RFI), as well as
signal loss due to capacitance (the tendency
of a nonconductor to store up electrical
charge). The performance of a twisted-pair
cable can be influenced by changing the
number of twists per foot in a wire pair.
IBM has developed its own categorization
system for twisted-pair cable, mainly to
describe the cable supported for IBM's
Token Ring network architecture. The sys-
tem is discussed in the Cable, IBM article.
A twisted-pair cable has the following
components:
Conductor wires: The signal wires for this
cable come in pairs that are wrapped
around each other. The conductor
wires are usually made of copper. They
may be solid (consisting of a single
wire) or stranded (consisting of many
thin wires wrapped tightly together).
A twisted-pair cable usually contains
multiple twisted-pairs; 2, 4, 6, 8, 25,
50, or 100 twisted-pair bundles are
common. For network applications,
2- and 4-pair cables are most com-
monly used.
Shield: Shielded twisted-pair (STP) cable
includes a foil shield around each pair
of conductors.
Jacket: The wire bundles are encased in
a jacket made of polyvinylchloride
(PVC) or, in plenum cables, of a
fire-resistant material, such as Teflon
or Kynar.
The figure "Components of twisted-pair
cable" shows the makeup of this type of
cable. Note that the shield is not included
for unshielded twisted-pair cable.
Twisted-pair cable comes in two main
varieties: shielded (STP) and unshielded
Twisted-Pair Cable Components
COMPONENTS OF
TWISTED-PAIR CABLE


Prev Page 174 Next

162
Cable, Twisted-Pair
(UTP). STP contains an extra shield or
protective screen around each of the wire
pairs to cut down on extraneous signals.
This added protection also makes STP more
expensive than UTP. (The price of coaxial
cable actually lies between UTP and STP
prices.)
STP cable has pairs of conductors twisted
around each other. Each pair is covered with
a foil shield to reduce interference and mini-
mize crosstalk between wire pairs.
STP can handle high-speed transmissions,
but the cable itself is relatively expensive,
can be quite bulky and heavy, and is rather
difficult to work with.
STP is used in ARCnet and Token Ring
networks, although the special cable ver-
sions developed by IBM are more likely
to be used in the Token Ring networks.
Several of the types specified in the IBM
Cable System are STP: Types 1, 2, 6, 8,
and 9 (see the Cable, IBM article).
UTP cable does not include any extra shield-
ing around the wire pairs. This type of cable
is used in some Token Ring networks, usu-
ally those working at slower speeds. UTP
can also be used in Ethernet and ARCnet
architectures.
UTP is not the primary choice for any
network architecture, but the IEEE has
approved a standard for a 10BaseT Ethernet
network that uses UTP cabling at 10 Mbps.
Networking mavens are divided as to
whether 10BaseT and the use of UTP
cable in general are welcome additions
or dead-ends.
Shielded Twisted-Pair (STP) Cable
Unshielded Twisted-Pair (UTP) Cable
Because it lacks shielding, UTP is not as
good at blocking noise and interference as
STP or coaxial cable. Consequently, UTP
cable segments must be shorter than when
using other types of cable. For standard
UTP, the length of a segment should never
exceed 100 meters (about 330 feet).
On the other hand, UTP is quite inexpen-
sive, and is very easy to install and work
with. The price and ease of installation make
UTP tempting, but keep in mind that instal-
lation is generally the major part of the
cabling expense (so saving on the cable
won't necessarily help cut expenses very
much) and that other types of cable may
be just as easy to install.
To distinguish varieties of UTP, the
EIA/TIA has formulated five categories.
These are summarized in the Cable Stan-
dards article.
Twisted-pair cable is described in terms of
its electrical and performance properties.
The features that characterize UTP and STP
cable include the following:
Attenuation: This value indicates how
much power the signal has lost and is
dependent on the frequency of the
transmission. Attenuation is measured
in relation to a specified distance; for
example, 100 meters, 1000 feet, or 1
kilometer. Attenuation per 1000 feet
values range from under 10 dB (for
Category 4 cable running at 1 MHz) to
more than 60 dB (for Category 5 cable
running at 100 MHz). With attenua-
tion, a lower value is better.
Performance Features


Prev Page 175 Next

Cable, Twisted-Pair
163
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
Capacitance: This value indicates the
extent to which the cable stores up
charge (which can distort the signal).
Capacitance is measured in picofarads
(pF) per foot; lower values indicate
USING EXISTING TELEPHONE
CABLE WIRES
Most telephone cable is UTP, and many tele-
phone cables have extra wires because the cable
comes with four pairs and the telephone com-
pany needs only two of the pairs for your tele-
phone connection. (Any additional lines or
intercoms require their own wire pairs.)
If there are unused wire pairs, you may be able to
use these for your network cabling. While this is
a tempting possibility, consider the following
points carefully:
I
The cable might not run conveniently for
your needs, so you may need to add cable
segments.
I
Make sure you test all the cable you'll be
using, and don't be surprised if some of it is
defective.
I
The telephone cable may be the lower-quality,
voice-grade type, and you really should be
using data-grade cable, unless you're transmit-
ting over very short distances.
If you're going to use already installed cable for
your network, make sure all of it works properly.
Use a cable tester, which can provide detailed
information about the cable's physical and electri-
cal properties. When you're dealing with a long
cable system, the chances are good that at least
parts of it will be faulty. Find and replace the bad
cable before you set everything up.
better performance. Typical values are
between 15 and 25 pF/ft.
Impedance: All UTP cable should have an
impedance of 100 +/- 15 ohms.
NEXT: The near-end crosstalk (NEXT)
indicates the degree of interference
from neighboring wire pairs. This is
also measured in decibels per unit dis-
tance, but because of notation and
expression conventions, a high value is
better for this feature. NEXT depends
on the signal frequency and cable cate-
gory. Performance is better at lower
frequencies and for cables in the higher
categories.
Twisted-pair cable has the following
advantages over other types of cables
for networks:
I It is easy to connect devices to twisted-
pair cable.
I If an already installed cable system,
such as telephone cable, has extra,
unused wires, you may be able to use
a pair of wires from that system. For
example, in order to use the telephone
cable system, you need telephone cable
that has four pairs of wires, and there
can be no intercoms or second lines to
use the two pairs not needed for the
telephone connection.
I STP does a good job of blocking
interference.
I UTP is quite inexpensive.
I UTP is very easy to install.
Twisted-Pair Cable Advantages


Prev Page 176 Next

164
Cable, Voice-Grade
I UTP may already be installed (but
make sure it all works properly and
that it meets the performance specifi-
cations your network requires).
Twisted-pair cable has the following dis-
advantages compared with other types of
cable:
I STP is bulky and difficult to work
with.
I UTP is more susceptible to noise and
interference than coaxial or fiber-optic
cable.
I UTP signals cannot go as far as they
can with other cable types before they
need cleaning and boosting.
I A skin effect can increase attenuation.
This occurs when transmitting data at
a fast rate over twisted-pair wire.
Under these conditions, the current
tends to flow mostly on the outside
surface of the wire. This greatly
decreases the cross-section of the wire
being used for moving electrons, and
thereby increases resistance. This, in
turn, increases signal attenuation,
or loss.
When you are deciding on a category of
cable for your needs, take future develop-
ments-in your network and also in technol-
ogy-into account. It is a good idea to buy
the cable at least one category above the one M
you have selected. (If you selected Category
5 cable to begin with, you should seriously
consider fiber-optic cable.)
Twisted-Pair Cable Disadvantages
Selecting and Installing Twisted-Pair Cable
Check the wiring sequence before you
purchase cable. Different wiring sequences
can lurk behind the same modular plug in a
twisted-pair cable. (A wiring sequence, or
wiring scheme, describes how wires are
paired up and which locations each wire
occupies in the plug.) If you connect a plug
that terminates one wiring scheme into a
jack that continues with a different
sequence, the connection may not provide
reliable transmission. See the Wiring
Sequence article for more information.
You should find out which wiring scheme
is used before buying cable, and buy only
cable that uses the same wiring scheme. If
you are stuck with existing cable that uses
an incompatible wiring scheme, you can use
a cross wye as an adapter between the two
schemes.
If any of your cable purchases include
patch cables (for example, to connect a com-
puter to a wallplate), be aware that these
cables come in two versions: straight
through or reversed. For networking appli-
cations, use straight-through cable, which
means that wire 1 coming in connects to
wire 1 going out (rather than to wire 8 as in
a reversed cable), wire 2 connects to wire 2
(rather than to wire 7), and so on. The tools
for installing and testing twisted-pair cable
are the same as those used generally for net-
work cables. (See the Cable article for a dis-
cussion of cable tools.)
SEE ALSO
Cable; Cable, Coaxial; Cable, Fiber-Optic
Cable, Voice-Grade
Old-time, unshielded twisted-pair, tele-
phone cable; category 1 in the EIA/TIA-568


Prev Page 177 Next

Cache Buffer Pool
165
a
b
c
d
e
f
g
h
i
j
kl
m
n
o
p
q
rstuvwxyz
specifications. This cable is suited to the
transmission of voice signals. Officially,
such cable is not considered suitable for
data-grade transmissions. In practice, it
generally works fine at low speeds, over
short distances, and under ordinary working
conditions.
SEE ALSO
Cable, Twisted-Pair
M
Cache
As a noun, a cache, also known as a disk
cache, is an area of RAM (random-access
memory) set aside for holding data that is
likely to be used again. By keeping fre-
quently used data in fast RAM, instead of
on a hard or floppy disk with much slower
access, a system's performance can be
improved greatly.
As a verb, cache refers to the process of
putting information into a cache for faster
retrieval. Directory information and hard
disk contents are examples of data likely to
be cached. The figure "Disk cache" shows
an example of this process.
M
Cache Buffer Pool
In Novell's NetWare, the cache buffer pool
is the amount of memory available for the
network operating system (NOS) after the
server module has been loaded into memory.
DISK CACHE


Prev Page 178 Next

166
Call
The memory in this pool can allocated for
various purposes:
I To cache the file allocation tables
(FATs) for each NetWare volume
I To create a hash table containing
directory information
I To provide memory for NetWare
Loadable Modules (NLMs) that are
needed
MCall
A request from one program or node to
begin a communication with another node.
The term is also used to refer to the resulting
communications session.
M
Caller ID
In ISDN and some other telecommunica-
tions environments, a feature that includes
the sender's identification number (such as
telephone number) in the transmission so
that the receiver knows who is calling.
Caller ID is also known as ANI (automatic
number identification) and CLID (calling
line identification).
MCall Setup Time
The amount of time needed to establish a
connection between two nodes so they can
communicate with each other.
M
Campus Area Network (CAN)
A network that connects nodes (or possibly
departmental local-area networks) from
multiple locations, which may be separated
by a considerable distance. Unlike a wide-
area network, however, a campus network
does not require remote communications
facilities, such as modems and telephones.
M
Campus-Wide Information System
(CWIS)
SEE
CWIS (Campus-Wide Information
System)
MCapacitance
Capacitance is the ability of a dielectric
(nonconductive) material to store electricity
and to resist changes in voltage. In the pres-
ence of a signal (a voltage change), the
dielectric will store some of the charge.
Capacitance is usually measured in micro-
farads or picofarads (millionths or trillionths
of a farad, respectively).
Other things being equal, the lower the
capacitance, the better the cable. A higher
capacitance means that more of the charge
can be stored in the dielectric between two
conductors, which means greater resistance.
At higher frequencies, high capacitance
results in greater signal attenuation.
SEE ALSO
Cable
M
Capacitor
An electrical component in line conditioners,
surge protectors, and other equipment.
Capacitors help clean incoming power by
absorbing surges and noise from electromag-
netic and radio frequency interference. Com-
pare it with inductor and MOV (metal oxide
varistor).


Prev Page 179 Next

CAT (Common Authentication Technology)
167
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
M
Carrier Band
A communications system in which the
entire bandwidth is used for a single trans-
mission and in which a signal is modulated
before being transmitted. This is in contrast
to baseband systems, which do not modu-
late the signal, and to broadband systems,
which divide the total bandwidth into multi-
ple channels.
M
Carrier Frequency
The rate at which the carrier signal repeats,
measured in cycles per second, or hertz. In
communications, the carrier signal is modu-
lated, or altered, by superimposing a second M
signal, which represents the information
being transmitted. In an acoustic signal, the
frequency represents the signal's pitch.
MCarrier On
In carrier sense, multiple access (CSMA)
media-access methods, a signal that indi-
cates the network is being used for a trans-
mission. When a node detects this signal, the
node waits a random amount of time before
trying again to access the network.
MCarrier Pulse
A signal, consisting of a series of rapid, con-
stant pulses, used as the basis for pulse mod-
ulation; for example, when converting an
analog signal into digital form.
M
Carrier Signal
An electrical signal that is used as the
basis for a transmission. This signal has
well-defined properties, but conveys no
information (content). Information is sent
by modifying (modulating) some feature of
the carrier signal, such as the amplitude, fre-
quency, or timing, to represent the values
being transmitted.
M
Carrier Wire
A conductive wire (capable of carrying an
electrical signal); for example, the central
wire in a coaxial cable, which serves as the
medium for the electrical signal.
SEE ALSO
Cable
CAS (Communicating Application
Specification)
An interface standard for fax modems devel-
oped by Intel and DCA. This proposed stan-
dard competes with the Class x hierarchy
developed by EIA.
M
CAT (Common Authentication
Technology)
In the Internet community, CAT is a specifi-
cation for distributed authentication under
development. CAT supports authentication
measures based on either public- or private-
key encryption strategies.
With CAT, both client and server pro-
grams must use the services of a common
interface, which will provide the authentica-
tion services. This interface will connect to
either DASS (Distributed Authentication
Security Service), which uses public-key
encryption, or Kerberos, which uses private-
key encryption.


Prev Page 180 Next

168
CAU (Controlled Access Unit)
BROADER CATEGOR Y
Authentication
SEE ALSO
DASS (Distributed Authentication
Security Service); Kerberos
MCAU (Controlled Access Unit)
In IBM Token Ring networks, the term for
an intelligent hub. CAUs can determine
whether nodes are operating, connect and
disconnect nodes, monitor node activity, and
pass data to the LAN Network Manager
program.
M
CAU/LAM (Controlled Access Unit/
Lobe Attachment Module)
In IBM Token Ring networks, a hub (the
CAU) containing one or more boxes
(the LAM) with multiple ports to which
new nodes can be attached.
MCBC (Cipher Block Chaining)
An operating mode for the DES.
SEE
DES (Data Encryption Standard)
MCBEMA (Computer Business
Manufacturers Association)
An organization that provides technical
committees for work being done by other
organizations; for example, the committee
for the FDDI standard published by ANSI.
M
CBMS (Computer-Based Messaging
System)
An older term for a Message Handling
System (MHS), or for electronic mail.
SEE
E-Mail
M
CBR (Constant Bit Rate)
An ATM connection that uses Class A ser-
vice, which is designed for voice or other
data that are transmitted at a constant rate.
Compare it with VBR (variable bit rate).
M
CC (Clearing Center)
In EDI, a message-switching element
through which documents are passed on
the way to their destinations.
SEE ALSO
EDI (Electronic Document Interchange)
MCCIR (International Consultative
Committee for Radiocommunication)
An ITU (International Telecommunication
Union) agency that is responsible for defin-
ing standards for radio communications.
In 1993, the CCIR-together with the IFRB
(International Frequency Registration
Board)-was replaced by the ITU-R
(International Telecommunication Union-
Radiocommunication Standardization
Sector).
SEE ALSO
ITU


Prev Page 181 Next

CCS (Common Communications Support)
169
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
M
CCIS (Common Channel
Interoffice Signaling)
In telephone communications, a transmis-
sion method that uses different channels for
voice and control signals. The control sig-
nals are sent by a fast, packet-switched
method, which makes it possible to include
extra information (such as caller ID and bill- M
ing information) in the control channel.
SEE ALSO
CCS 7
M
CCITT (Consultative Committee
for International Telegraphy and
Telephony)
The CCITT is a permanent subcommittee of
the ITU (International Telecommunications
Union), which operates under the auspices
of the United Nations. The committee con-
sists of representatives from 160 member
nations, mostly from national PTT (Postal,
Telephone, and Telegraph) services.
The CCITT is responsible for dozens of
standards used in communications, telecom-
munications, and networking, including the
X.25 and X.400 standards, the V.42 and
V.42bis standards for modems, and the
I.xxx series of documents on ISDN (Inte-
grated Services Digital Network).
The CCITT works closely with the
ISO (International Standardization
Organization), so that many standards and
recommendations will appear in documents
from both groups. CCITT recommendations
appear every four years, with 1992 (the
white books) being the most recent.
In March 1993, the CCITT was officially
renamed the International Telecommunica-
tion Union-Telecommunication Standardi-
zation Sector (ITU-T, sometimes written as
ITU-TS or ITU-TSS). However, since the
CCITT name is so familiar and is likely to
remain in widespread use for some time, the
older name is used throughout this book.
CCRSE (Commitment, Concurrency,
and Recovery Service Element)
In the OSI Reference Model, an application-
layer service that is used to implement
distributed transactions among multiple
applications.
SEE
ASE (Application Service Element)
M
CCS (Common Channel Signaling)
A signaling method in which control signals
are sent across different channels than voice
and data signals. This makes it possible to
include various types of extra information in
the control signal.
SEE ALSO
CCS 7
MCCS (Common Communications
Support)
One of the pillars of IBM's SAA specifica-
tions. CCS includes support for data links,
application services, session services, and
data streams.
SEE ALSO
SAA (Systems Application Architecture)


Prev Page 182 Next

170
CCS (Continuous Composite Servo)
M
CCS (Continuous Composite Servo)
A compact disc recording technique in
which the contents are stored on separate
tracks laid out in concentric circles.
COMPARE
SS (Sampled Servo)
M
CCS (Hundreds of Call Seconds)
In telephone communications, a measure of
line activity. One CCS is equivalent to 100
seconds of conversation on a line, so that an
hour of line usage is 36 CCS; 36 CCS is
equal to one Erlang, and indicates continu-
ous use of the line.
M
CCS 7 (Common Channel Signaling 7)
A version of the CCITT's Signaling System 7
(SS7); a transmission method in ISDN that
makes special services (such as call forward-
ing or call waiting) available anywhere in
a network. CCS 7 is an extension of the
CCIS method for transmitting control
information.
MCD (Carrier Detect)
A signal sent from a modem to a PC, to indi-
cate that the modem is on line and ready for
work.
M
CD (Compact Disc)
Compact discs are the product of a record-
ing and storage technology that makes it
possible to fit over half a gigabyte of digital
data on a disc about the size of a floppy
disk. Unlike floppy or hard disks, which use
magnetic technology, compact discs are
recorded using optical methods.
To produce a master disc for commer-
cially produced CDs, a laser literally burns
the information into the disc by creating tiny
pits in the surface. This changes the reflec-
tive properties of the disc at these locations
relative to the surrounding surface. The
information is read by using a laser so that
there is never any physical contact during
the reading process. The information on a
CD is actually contained in the transitions
between the pits and the non-pit areas
(known as the lands).
CD technology has undergone several
revisions and advancements since the first
digital audio (DA) discs were developed over
10 years ago.
The following standards and variants have
been created and used over the years. Most
of these standards are still in use, and many
current CD drives can read several of the
standards. In addition, newer standards
(such as CD-XA) are often back-compatible
with earlier standards (such as CD-ROM).
CD standards are distinguished by the
color of the laser used in that particular
technology-for example, red, yellow, and
green. Collectively, these standards docu-
ments are known as the Rainbow Books.
The following standards are among the most
popular:
CD-DA (Digital Audio) (Red Book) This
was the first compact disc standard,
and was developed for recording musi-
cal discs. CD-DA discs can hold about
74 minutes of music recorded at
CD Variants


Prev Page 183 Next

CD (Compact Disc)
171
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
44,100 samples per second (known as
the scanning frequency), using PCM
(pulse code modulation) as the digiti-
zation method, and allocating 16 bits
for each sample. (With 16 bits, each
sample can take on any of 65, 536 (or
216 values). These bits can be allocated
in whatever manner one chooses, pro-
vided the resulting split is meaningful.
For example, by allocating 8 bits
to each channel, you can get stereo.
CD-DA was not developed for record-
ing data. CD-DA is what everyone cor-
rectly thinks of as audio CD.
CD-ROM (Read Only Memory) (Yellow
Book) This standard was designed
to enable CD technology to be used
with computers-and for storing huge
amounts of data. Because error rate
requirements for data are much more
stringent than for music, the bits in a
CD-ROM sector are allocated differ-
ently than for a musical performance.
Whereas a CD-DA sector has 2352
bytes available for storing music in
each sector, CD-ROM has only 2048,
because 280 extra bits had to be allo-
cated for error-detection and correc-
tion. CD-ROM actually does have a
less stringent mode, known as mode
2 (in contrast to the mode 1 used
for data). This makes 2336 bytes
per sector available for use (at the
cost of a considerable amount of
error correcting).
CD-ROM/XA (Extended Architecture)
(Yellow Book and some of the Green
Book) This standard was designed to
provide a more efficient and flexible
storage method, but one that could be
made back-compatible with earlier
standards. In addition to providing a
new, more flexible sector format, CD-
ROM/XA uses a different digitization
method and compresses the audio
data-decompressing the audio on the
fly if the audio should ever be needed.
At its lowest scanning frequency and
highest compression, a CD-ROM/XA
disc can hold over nine hours of stereo
music-compared to just under 1.25
hours for CD-DA. In addition, CD-
ROM/XA uses a new sector format,
which allows a file to be nested inside
another. Even though it uses special
hardware, CD-ROM/XA technology
is back compatible with CD-DA and
ordinary CD-ROM. (Fortunately, most
CD drives available today include this
extra hardware, so that these drives
can read most kinds of CDs.) CD-
ROM discs can hold up to 660 MBytes
of data.
Photo-CD This disc format was created
by Kodak to provide a way for cus-
tomers to digitize their photos and to
use them at work or home. The Photo-
CD technology combines the XA stan-
dards with multisession technology. A
session is a recording period. Origi-
nally, CD's could record only once,
which meant that all data or pictures
had to be recorded in a single session.
With a multisession disc, on the other
hand, a customer can have pictures
recorded several times up to the disc's
capacity.


Prev Page 184 Next

172
CD (Compact Disc)
CD-WO (Write Once) and CD-MO
(Magneto-optical) (Orange Book)
These are specifications for recordable
CDs. CD-WO-also known as CD-
WORM (Write once, read many)-
is the older standard. It can create
discs with capacities of 128 Mbytes,
650 Mbytes, or 6.5 GBytes, depending
on the disc's size. CD-WO discs
require a magneto-optical drive and
are not compatible with CD-ROM
technology. CD-MO discs can hold
128-, 230-, 600-, 650-, or 1300
MBytes, and they must also be read by
a special magneto-optical drive. Unlike
CD-WO, however, CD-MO discs can
be recorded multiple times. Because of
this, MO discs are also known as EO
(erasable optical) discs.
CD-R (Recordable) (Orange Book) This
is a variant of the WO standard.
Unlike CD-WO, however, discs
recorded using CD-R technology can
be read on ordinary CD-ROM drives.
Until recently, CD-R machines were
much too expensive for personal use;
this has begun to change, and such
devices are becoming very popular for
business use. Discs for use in a CD-R
drive are distinguished by their gold
surface, as opposed to the silvery sur-
face of a commercially produced disc.
One reason for this is that CD-R discs
are created using a somewhat different
process than commercial CDs. Instead
of burning pits into the surface, the
recording laser in a CD-R drive simply
changes the optical properties of an
organic paint in the disc's recording
surface. This makes it possible to work
with a much weaker laser. CD-R discs
can hold up to 660 MBytes of infor-
mation. These discs are, in essence, just
ordinary CD-ROM discs produced by
special means.
CD-I (Interactive) (Green Book) This
standard allows branching based on
interaction between the user and the
material. CD-I drives connect to a tele-
vision set. Any computing capabilities
required to run the software are built
into the drive. You cannot use or even
read CD-I discs in ordinary CD-ROM
drives. 3DO is a proprietary variant of
the CD-I standard.
High density CD (Blue Book) This tech-
nology is still being developed. When
perfected, this standard is expected to
increase the capacity of a disc ten-
fold-to about 6.5 GBytes. Look for
this technology in the next year or so.
Hybrid standards Several variants have
been developed for special purposes or
to make use of particular technology.
In general, such discs require special
hardware. Hybrids include CD+G,
CD-MIDI, CD-EB, and CD-V. CD+G
(for graphics) is basically an audio CD
with additional information such as
text or graphics. CD-MIDI (for Musi-
cal Instrument Digital Interface) is an
audio disc with MIDI information.
CD-EB (for Electronic Book) is special
size and format that is used mainly to
store reference materials. CD-V (for
video) is an audio disc with video
information recorded in analog form.


Prev Page 185 Next

CDPD (Cellular Digital Packet Data)
173
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
The laserdisc is actually a CD-V
variant.
The logical structure of the material on a
CD is defined in the ISO 9660 documents.
These, in turn, are based on the earlier High
Sierra specifications.
M
CDDI (Copper Distributed Data
Interface)
A networking configuration that implements
the FDDI architecture and protocols on
unshielded twisted-pair (UTP) cable-that
is, on electrical (rather than optical) cable. A
related implementation is SDDI (shielded
distributed data interface), which uses
shielded twisted-pair (STP) cable. Also
known as copper-stranded distributed data
interface and as TPDDI (twisted-pair DDI).
SEE ALSO
FDDI (Fiber Distributed Data Interface)
MCDFS (CD-ROM File System)
A file structure used for storing information
on a compact disc. The file allocation table
(FAT) system may not be efficient or even
feasible for such a disc because of the large
number of files the disc may contain.
M
CDMA (Code Division Multiple Access)
In cellular communications, CDMA is a pro-
posed transmission method that uses special
codes to fit up to ten times as much informa-
tion into a channel. Each signal that comes
in on a given frequency is "spread" using a
different code. When the receiver decodes
the received signals, only the signal with the
appropriate spread will be meaningful; the
other signals will be received as noise.
CDMA uses a soft-handoff when switch-
ing a transmission from one cell to another
to ensure that no bits are lost in the trans-
mission. In this type of handoff, both cells
transmit the transitional bits at the same
time and on the same frequency. This way,
one of the transmissions will be within range
of the receiver.
This method is not compatible with the
TDMA (time division multiple access)
method that was adopted as a standard in
1989.
BROADER CATEGOR Y
Cellular Communications
COMPARE
TDMA (Time Division Multiple Access)
M
CDPD (Cellular Digital Packet Data)
A cellular communications technology that
sends digital data over unused cellular
(voice) channels. CDPD data can be trans-
mitted at 19.2 kbps, but only in service areas
that support CDPD. Currently only a few
dozen of the major service areas around the
country provide direct CDPD support.
CDPD can be used as a mobile computing
strategy to stay connected with the company
network back at the office. Essentially, a
mobile user needs a special CDPD modem
and the appropriate software. The user gets
an IP (Internet protocol) address, which
makes it possible to communicate as well as
to make use of Internet services.
Mobile users can remain connected even
when they are not using their computers and
even when they are outside the range of a


Prev Page 186 Next

174
CD-ROM Drive
cell that supports CDPD. The CDPD specifi-
cations support a "sleep" mode for the com-
puter. The network signals periodically to
sleeping devices, and a device will "wake"
if the signal includes the device's name or
address. The monitoring for each device is
done by the MDIS (mobile data intermediate
system).
Similarly, the MDIS allows a user to
remain connected even beyond areas that
support CDPD through a technology known
as switched CDPD. If the user is outside a
service area with CDPD capabilities when
called, the MDIS opens a circuit-switched
connection over the channel. The connection
is circuit-switched as far as the cellular
network is concerned, but is essentially
packet-switched as far as the device is con-
cerned. This is because the MDIS closes the
connection whenever there is silence, and
reopens it whenever there is activity.
CDPD supports data compression and
encryption. This cuts down on transmission
times (and costs) and also helps keep snoop-
ers from getting access to the data. In the
CDPD specification, the data are first
compressed and then encrypted.
The CDPD specification is being
formulated under the auspices of the
CDPD Forum, which you can contact at
info@forum.cdpd.net or at 800-335-CDPD
(2373).
SEE ALSO
Cellular Communications
M
CD-ROM Drive
CD-ROM stands for compact-disc,
read-only memory. A CD-ROM drive
is a peripheral device for reading CDs,
which have a huge capacity (660
megabytes).
Several features distinguish CD-ROM
drives from each other:
I
Transfer rate, which represents the
amount of data that the drive can read
from the disc in a second. Speeds are
based on a base rate of 150 kbytes
per second, which is known as a
single-speed drive. Double speed and
quad-speed drives can transfer 300
and 600 kbytes per second, respec-
tively. Quad speed drives are the
current norm, but 6x drives (not
yet known as "hex speed") are also
available.
I Access time, which represents the aver-
age time it takes to find a specified item
of information on the disc. Currently,
access times of less than 200 msec are
considered standard.
I Compatibility with various CD
standards, which indicates the types
of CDs the drive can read. The CD
(compact disc) article summarizes
these. Briefly, drives should be able to
read CD-XA (extended architecture)
discs and should support multisession
formats.
I Number of discs the drive can handle.
Multidisc systems can hold 3, 6, or
even 18 discs, and can switch between
them within a few seconds. The drive
can only read one disc at a time,
however.
A CD-ROM drive may be connected
to a network, making any available CDs


Prev Page 187 Next

Cell Loss Priority
175
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
shareable resources. With the appropriate
server and drivers, users can share access to
the disc currently loaded in the drive. A CD-
ROM drive can be accessed just like any
other volume, except that you can only read
from it. If there are licensing restrictions on
the use of a disc, it is essential that the server
software be able to restrict simultaneous
access to the licensed number of users.
Like any other type of hardware device,
CD-ROM drives require hardware drivers to
communicate. In addition, a special driver
containing extensions is required. These
extensions are specific to the operating
system, such as DOS, OS/2, or NT, with
which the CD-ROM is working. Microsoft
has provided such a driver for MS DOS,
called MSCDEX, which can be used with
most CD-ROM drives. Some hardware
manufacturers have also created their own
proprietary drivers. If you are connecting a
CD-ROM drive to a workstation, you will
need to load both the driver's regular hard-
ware driver and either MSCDEX or the
manufacturer's own extensions driver.
If you want to make a CD-ROM drive
available as a shared volume on a NetWare
3.12 or NetWare 4.x network, you do not
load the MSCDEX driver. Instead, load the
CD-ROM driver's regular hardware drivers
and Novell's CDROM.NLM. This NLM
manages the interface between the drive and
NetWare and enables the CD-ROM device
to be viewed and accessed by multiple users,
just like any other NetWare volume.
Note that the drivers available for a given
CD-ROM drive may or may not work with
your system. Verify that the drive is compat-
ible before you install it.
MCD-ROM File System (CDFS)
SEE
CDFS (CD-ROM File System)
MCell
In communications or networking, a packet,
or frame, of fixed size. In general, fast
packet-switching technologies-such as
ATM (Asynchronous Transfer Mode) and
SDMS (Switched Multimegabit Digital Ser-
vice)-use cells. Slower packet-switching
technologies-such as X.25-are more
likely to use variable-sized packets.
In cellular communications, a cell refers
to a geographic area. Each cell has its own
transmitter and receiver, through which sig-
nals can be distributed throughout the cell.
Transmissions must be "handed off" from
one cell to another when a mobile telephone
or networking caller actually moves from
one cell to another.
M
Cell, ATM
In the broadband ATM (Asynchronous
Transfer Mode) network architecture, cell
refers to a packet. ATM cells are each 53
octets, of which five octets are header and
48 are data.
SEE ALSO
ATM (Asynchronous Transfer Mode)
MCell Loss Priority
In an ATM network, a bit value that
specifies whether a cell can be discarded
if advisable; for example, if the network


Prev Page 188 Next

176
Cellular Communications
gets too busy. A value of 1 indicates an
expendable cell.
SEE ALSO
ATM (Asynchronous Transfer Mode)
M
Cellular Communications
Cellular communications is a wireless com-
munications technology. The communica-
tions area is divided into smaller areas,
called cells, and transmissions are passed
from cell to cell until they reach their desti-
nations. Each cell contains an antenna and
transmission facilities to pick up signals
from another cell or from a caller and to
pass them on to an adjacent cell or to a
callee within the cell. Cells can be anywhere
from a few kilometers to 32 kilometers
(20 miles) in diameter.
One cellular communications method,
called CDPD (Cellular Digital Packet Data)
transmits data over any cellular channels
that are not being used. CDPD uses tele-
phone (voice) channels, but can switch to a
new frequency, if necessary, when a voice
transmission begins in the cell being used.
CDPD was developed to provide data com-
munications in the cellular frequency range
without interfering with voice calls.
MCELP (Code Excited Linear Predictive M
Coding)
A variant of the LPC voice encoding algo-
rithm. CELP can produce digitized voice
output at 4,800 bits per second.
SEE ALSO
LPC (Linear Predictive Coding)
M
Central Office
The telephone switching station nearest to a
customer (residential or business). Custom-
ers are connected directly to a CO, which
connects them to other points in the tele-
communications hierarchy. The CO pro-
vides services such as switching, dial tone,
private lines, and centrex.
M
Central Processing
Central processing, also known as central-
ized processing, is a network configuration
in which a single server processes tasks for
multiple stations, all of which can commu-
nicate with the server. In such a setup, the
nodes must share the computing power of
the central processor. One consequence is
that the more tasks, the slower things get
done.
Central processing can be compared with
distributed processing, in which tasks are
performed by specialized nodes somewhere
on a network. A station that needs some-
thing done sends a request onto the net-
work. The server responsible for the service
takes on the task, does it, and returns the
results to the station. The client station need
never know who actually did the work.
CERT (Computer Emergency
Response Team)
In the Internet community, CERT is a
group formed in 1988 (by DARPA) to help
respond to, and deal with security problems
that may arise on the Internet. The group
also provides Internet administrators with


Prev Page 189 Next

Character
177
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
information and assistance to help avoid
security problems.
Tools and documents related to network
security are available through Anonymous
FTP from CERT's database in cert.org.
See the Protocol, FTP article for more
information.
BROADER CATEGOR Y
Network Security
M
CFB (Cipher Feedback)
An operating mode for the DES.
SEE
DES (Data Encryption Standard)
M
CGI (Common Gateway Interface)
An interface specification that defines the
rules of communication between informa-
tion servers, such as HTTP (Hypertext
Transport Protocol) servers on the World
Wide Web and gateway programs. More
specifically, the CGI is used when such a
server needs to pass a user request to a gate-
way program. Being able to pass work off to
the gateway program helps take some of the
workload off the server.
The gateway program is generally
designed to provide a mechanism for getting
input from a user-for example, so an
authorized user can complete an authentica-
tion form in order to get access to restricted
areas. Among other things, the CGI specifi-
cations define the mechanisms by which
information can pass from the server to the
gateway program and back.
The CGI specifications, along with many
of the other specifications related to HTTP
environments, are still undergoing revisions.
PRIMAR Y SOURCES
You can find the current form of the CGI
specifications at
http://hoohoo.ncsa.uiuc.edu.cgi/
overview.html
M
Channel
A channel is a physical or logical path for a
signal transmission. Two particularly impor-
tant channels in networking are the commu-
nications channel and the disk channel.
A communications channel is a path
through which data or voice can be trans-
mitted; for example, in a network or a
telephone call. In telecommunications, a
single cable may be able to provide multiple
channels.
A disk channel, in a hard-disk configura-
tion, consists of the components that con-
nect a hard disk drive to an operating
environment, such as DOS, OS/2, NetWare,
or VINES. These components include cables
and a hard disk adapter or controller. A sin-
gle channel can accommodate multiple hard
disks. A computer may have multiple disk
channels.
COMPARE
Circuit
MChannel Bank
A device that multiplexes low-speed signals
into a single high-speed signal.
M
Character
A byte with an identity. A group of bits, usu-
ally, seven or eight bits, that represents a sin-
gle letter, digit, special symbol, or control


Prev Page 190 Next

178
Checksum
code in an encoding scheme, such as ASCII
or EBCDIC.
MChecksum
Checksum is a simple error-detection strat-
egy that computes a running total based on
the byte values transmitted in a packet, and
then applies a simple operation to compute
the checksum value.
Checksums are very fast and easy to
implement, and they can detect about
99.6 percent of errors in a packet. This
reliability level is acceptable for most sim-
ple communications situations, but is less
reliable than the more sophisticated CRC
(Cyclical Redundancy Check) calculations,
which have an accuracy of more than 99.9
percent.
The receiver compares the checksums
computed by the sender and by the receiver.
If they match, the receiver assumes the trans-
mission was error-free. If they do not match,
there was an error.
BROADER CATEGOR Y
Error Detection and Correction
COMPARE
CRC (Cyclical Redundancy Check);
Parity
MChromatic Dispersion
In a fiber-optic transmission, the dispersion
of a light signal because of the different
propagation speeds of the light at different
wavelengths; also known as material disper-
sion. The wavelengths around which disper-
sion is minimal, such as those around 1300
or 830 nanometers, are commonly used for
signaling.
M
CHRP (Common Hardware Reference
Platform)
A set of specifications for PowerPC systems.
CHRP is being developed by Apple, IBM,
and Motorola, and is designed to enable
such a machine to run multiple operating
systems and cross-platform applications.
While specifications have not been final-
ized, a minimum machine will have at least
8 MB of RAM and a 1 MB cache; CHRP
machines will use the PowerPC 604 or later
chip, and will support the PCI (Peripheral
Component Interconnect) bus standard.
CHRP machines will support at least the
following environments:
I AIX (IBM's UNIX port)
I IBM OS/2 for PowerPC
I Mac OS (Apple's new Macintosh
operating system)
I Novell NetWare
I Solaris (from SunSoft)
I Microsoft Windows NT
M
CICS (Customer Information Control
System)
A terminal that provides transaction pro-
cessing capabilities for IBM mainframes.
CICS supports the SNA (Systems Network
Architecture).
M
CIDR (Classless Interdomain Routing)
CIDR is a routing strategy that was devel-
oped as a partial solution to two difficulties
that have developed as the number of


Prev Page 191 Next

CIDR (Classless Interdomain Routing)
179
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
networks connected to the Internet has
grown very large. One problem was that
routers had to deal with too many network
addresses and were choking on their routing
tables. The second problem was that the
supply of Class B network addresses was
being used up too quickly. Class B networks
can have up to 65,536 hosts, but there can
be only 16,384 Class B network addresses.
This address class (see IP Address for a more
detailed discussion) is useful for companies
or organizations that have large networks
with thousands of hosts for each network.
While there are many companies with a few
thousand hosts on their networks, there are
few that have anywhere near 65,000.
Because of the way address classes are
defined, this situation leads to a lot of poten-
tial addresses being wasted. The next
address class-C-supports networks with
256 or fewer hosts. There can be more than
2 million Class C addresses. So, whereas
Class B address spaces are too big, those
for Class C are somewhat small for many
businesses and organizations. When a mid-
size company asks for an Internet address, it
must be given either a Class B address from
a dwindling supply or several (perhaps sev-
eral dozen) Class C addresses. For example,
a company with just over 8,000 hosts would
need 32 Class C addresses. In contrast, by
taking a Class B address, it would waste
more than 55,000 potential addresses.
CIDR is designed to make a happy
medium possible by assigning consecutive
Class C addresses to organizations or corpo-
rations that have more than 256 machines,
but that may not be large enough to merit a
Class B address. CIDR takes advantage of
the assignment scheme and treats the cluster
of Class C networks as belonging to the
same "supernetwork"-as indicated by their
common value in the higher order address
bits (known as the prefix bits in this con-
text). By routing just on the (fewer) higher-
order bits, routers can fulfill their functions
without having to store all the networks to
which they are routing.
For CIDR to be successful, several things
are required:
I The internal and external gateway pro-
tocols need to be able to represent the
"supernetwork cluster" groupings.
Earlier gateway protocols (such as
BGP-3, IGRP, and RIP-1) cannot do
this; newer versions (such as BGP-4,
EIGRP, IS-IS, OSPF, and RIP-2) can.
The protocol situation is in transition
because newer protocols are, in some
cases, just becoming available.
I Class C addresses must be assigned
consecutively, as assumed in the CIDR
strategy. While this can be done easily
in some areas, it's much more difficult
in others. One important and sticky
issue is how to deal with address
owners who move, as such a move
could entail a switch in providers,
which would undoubtedly lead to
routing changes. If the address that's
moving happens to be in the middle of
a "supernetwork," the abbreviated
addressing scheme falls apart.
I An effective strategy must be worked
out for assigning addresses. Two possi-
ble basic approaches are provider-
based and geographically based. In the
former, networks that share a provider


Prev Page 192 Next

180
Cipher Block Chaining (CBC)
get addresses close to each other,
regardless of whether these networks
are physically near each other. The
geographically based approach would
assign addresses within a block to net-
works in the same geographical area.
The current Internet is closer to the
provider-based variant.
MCipher Block Chaining (CBC)
An operating mode for the DES.
SEE
DES (Data Encryption Standard)
MCipher Feedback (CFB)
An operating mode for the DES.
SEE
DES (Data Encryption Standard)
MCiphertext
Text that has been encrypted to make it
unintelligible to anyone who lacks essential
information about the encryption scheme.
The required information is generally a spe-
cific value, known as the encryption (or
decryption) key. Conventional-, public-,
or private-key encryption strategies may
be used to create ciphertext.
SEE ALSO
Plaintext
M
CIR (Committed Information Rate)
In frame-relay networks, a bandwidth, or
information rate, that represents the average
level for a user. If the user's network activity
exceeds this rate, the frame-relay controller
will mark the user's extra packets to indicate
that they can be discarded if necessary.
MCircuit
A closed path through which electricity
can flow. The term is also used to refer to
components (such as chips) capable of creat-
ing such a path.
MCIS (CompuServe Information
Service)
CIS, better known simply as CompuServe, is
the oldest of the major online services, and
is still the largest, although America Online,
or AOL, is gaining rapidly. CompuServe
supports DOS, Windows, and Macintosh
users. It offers the usual forums, electronic
mail, financial and news services, and soft-
ware to download or use online. For a flat
monthly fee, users have unlimited access to
basic services; special services incur addi-
tional fees. For a fee, users can also get
access to the Internet.
SEE ALSO
AOL (America Online); Prodigy
FOR INFORMATION
Call (800) 848-8199
MCISC (Complex Instruction Set
Computing)
CISC is a processor design strategy that pro-
vides the processor with a relatively large
number of basic instructions, many of which
are complex but very powerful. These com-
plex instructions may require several clock
cycles to complete, which can slow down
overall processing.


Prev Page 193 Next

Client
181
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
CISC is in contrast to the RISC (reduced
instruction set computing) design strategy. A
RISC chip uses a small number of sim-
ple operations to do its work. These simple
operations are optimized for speed, and
most require only a single clock cycle for
completion.
MCIX (Commercial Internet Exchange) M
CIX is an association of domestic Internet
access providers that provides connection
points between commercial traffic and the
Internet. The CIX was formed to route
commercial traffic back when such traffic
was not allowed according to the AUP
(acceptable use policy) for the Internet. CIX
members agree to carry each others' traffic
when requested. Contact Gopher or Web
servers at cix.org for more information
about CIX.
M
Cladding
In fiber-optic cable, the material (usually
plastic or glass) surrounding the fiber core.
The cladding has a lower index of refraction
than the core, which means that light hitting
the cladding will be reflected back into the
core to continue its path along the cable.
SEE ALSO
Cable, Fiber-Optic
M
Clamping Time
In power protection, the amount of time
needed for a surge protector to deal with
a voltage spike or surge; that is, to bring the
voltage within acceptable levels.
MClass A Certification
An FCC certification for computer or other
equipment intended for industrial, commer-
cial, or office use, rather than for personal
use at home. The Class A commercial certifi-
cation is less restrictive than the Class B
certification.
Class B Certification
An FCC certification for computer equip-
ment, including PCs, laptops, and portables
intended for use in the home rather than in a
commercial setting. Class B certification is
more restrictive than the commercial Class
A certification.
M
Clearing Center (CC)
SEE
CC (Clearing Center)
M
CLID (Calling Line Identification)
In ISDN and some other telecommunica-
tions environments, a feature that includes
the sender's identification number (such as
telephone number) in the transmission so
that the receiver knows who is calling. It
is also known as ANI (automatic number
identification) and caller ID.
M
Client
A client is a machine that makes requests of
other machines (servers) in a network or
that uses resources available through the
servers.
For example, workstations are network
clients because they use services from the


Prev Page 194 Next

182
Client-Based Application
server. As another example, a client applica-
tion is an application that makes requests
of other applications, on the same or on
different machines, for services, informa-
tion, or access to resources.
COMPARE
Server
SEE ALSO
Workstation
M
Client-Based Application
An application that executes on the client
machine (the workstation) in a network.
MClient/Server Computing
Client/server computing is a network-
ing arrangement with the following
characteristics:
I Intelligence, defined either as process-
ing capabilities or available informa-
tion, is distributed across multiple
machines.
I Certain machines-the clients-can
request services and information from
other machines-the servers. For
example, a server may have quick
access to huge databases that can be
searched on behalf of the client.
I The server does at least some of the
processing for the client.
Applications capable of running in a
client/server environment can be split into a
front end that runs on the client and a back
end that runs on the server. The front end
provides the user with an interface for giving
commands and making requests. The appli-
cation's real work is done by the back end,
which processes and carries out the user's
commands.
Client/server computing allows for sev-
eral types of relationships between the server
and client, including the following:
I Stand-alone (non-networked) client
applications which do not request
access to server resources. For
example, a local word processor might
be a stand-alone client application.
I Applications that run on the client but
request data from the server. For
example, a spreadsheet program might
run on a workstation and use files
stored on the server.
I Programs where the physical search of
records takes place on the server, while
a much smaller program running on
the client handles all user-interface
functions. For example, a database
application might run this way on the
server and client.
I Programs that use server capabilities to
share information between network
users. For example, an electronic-mail
system may use the server this way.
The figure "Client/server computing
arrangements" illustrates these different
arrangements.
SEE ALSO
Back End; Front End


Prev Page 195 Next

Clock Speed
183
a
b
c
d
efghijklmno
p
q
r
s
t
u
v
w
x
y
z
CLIENT/SER VER COMPUTING ARRANGEMENTS
M
CLNS (Connectionless Mode
Network Service)
In the OSI Reference Model, CLNS is a
network-layer service in which data
transmission can take place without a fixed
connection between source and destination.
Individual packets are independent, and they
may reach the destination through different
paths and in a mixed order. In this type of
transmission service, each packet must carry
its own destination address and information M
about the packet's relative position in the
message.
CLNS is the most common operating
mode for local-area networks (LANs). In
contrast, for wide-area networks (WANs),
CONS (connection-oriented network
service) is more popular.
PRIMAR Y SOURCE
ISO document 8348
BROADER CATEGOR Y
Connectionless Service
COMPARE
Connection-Oriented Service
Clock Speed
Activities carried out by and for the proces-
sor must all be carefully timed and coordi-
nated. To make this possible, each processor


Prev Page 196 Next

184
CLP (Cell Loss Priority)
has a clock associated with it. This clock
serves as a timing reference by slicing time
into very short intervals. The clock speed
is defined as the number of such slices in
a second.
Clock speed is expressed in millions of
cycles per second (megahertz, or MHz). For
example, the CPU in the original IBM had a
clock speed of 4.77 MHz. This is painfully
slow when compared to today's processors,
with clock speeds that can be in the 100
MHz range.
MCLP (Cell Loss Priority)
In an ATM network, a bit value that
specifies whether a cell can be discarded
if advisable; for example, if the network
gets too busy. A value of 1 indicates an
expendable cell.
SEE ALSO
ATM (Asynchronous Transfer Mode)
MCLTS (Connectionless Transport
Service)
In the OSI Reference Model, a transport-
layer service that does not guarantee deliv-
ery, but makes a best effort, does error
checking, and uses end-to-end addressing.
M
CLU (Command Line Utility)
In Novell's NetWare and in other operating
and networking environments, a program
that can be executed at the appropriate
command-line prompt. Examples of com-
mand line utilities in NetWare include
NCOPY and FLAG for manipulating
files and file attributes, respectively.
M
Cluster
In a network, particularly in a mainframe-
based network, a group of I/O (input/ouput)
devices, such as terminals, computers, or
printers, that share a common communica-
tion path to a host machine. Communica-
tions between the devices in a cluster and the
host are generally managed by a cluster con-
troller, such as IBM's 3274 controller.
MCluster Controller
A device that serves as an intermediary
between a host machine, such as a
mainframe, and a group (cluster) of I/O
(input/ouput) devices, such as terminals,
computers, or printers. The IBM 3274 is an
example of such a device. This controller has
been superseded by the 3174 establishment
controller.
MCMC (Common Mail Calls)
An API (Application Program Interface)
developed by the X.400 API Association
(XAPIA) to enable message-handling
agents-for example, in an email system-to
communicate with message stores, or post
offices. The calls in the API are designed
to be independent of hardware platforms,
operating systems, email systems, and mes-
saging protocols. The API is also referred
to as common messaging calls.
M
CMIP (Common Management
Information Protocol)
A network management protocol for the
OSI Reference Model. CMIP, pronounced
"see-mip," defines how management


Prev Page 197 Next

CMISE (Common Management Information Service Element)
185
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
information can be communicated between
stations. CMIP is functionally comparable
to the older, and arguably more widely used,
SNMP (Simple Network Management
Protocol).
SEE ALSO
Network Management
M
CMIPDU (Common Management
Information Protocol Data Unit)
In the OSI network management model,
a packet that conforms to the CMIP. The
packet's contents depend on the requests
from a CMISE, which relies on the CMIP to
deliver the user's requests and to return with
answers from the appropriate application or
agent.
SEE ALSO
CMISE (Common Management
Information Service Element);
Network Management
M
CMIPM (Common Management
Information Protocol Machine)
In the OSI network management model,
software that accepts operations from a
CMISE user and initiates the actions needed
to respond and sends valid CMIPDUs
(CMIP packets) to a CMISE user.
SEE ALSO
CMISE; Network Management
M
CMIS (Common Management
Information Service)
In the OSI network management model,
a standard for network monitoring and
control services. CMIS, pronounced "see-
miss," is documented in CCITT recommen-
dation X.710 and ISO document 9595.
SEE ALSO
CMISE; Network Management
MCMISE (Common Management
Information Service Element)
In the OSI network management model, a
CMISE is an entity that provides network
management and control services. Seven
types of CMISEs, pronounced "see-mize,"
are specified:
I
Event report
I Get
I Cancel get
I Set
I Action
I Delete
I Create
The services provided by CMISEs are
used by the system management functions
(SMFs). The SMFs are in turn used to carry
out the tasks specified for the five system
management functional areas (SMFAs)
defined in the OSI network management
model. The figure "Major components in
the ISO-OSI network management model"
shows this relationship.
SEE ALSO
Network Management


Prev Page 198 Next

186
CMOS (Complementary Metal-Oxide Semiconductor)
M
CMOS (Complementary Metal-Oxide
Semiconductor)
CMOS, pronounced "see-moss," is a logic
family for digital circuits. CMOS logic is not
exceptionally fast, but it has relatively low
power consumption, which makes it ideal
for such items as battery-powered PCs.
CMOS is used for RAM chips that
need to retain information, such as configu-
ration data or date and time information.
The values stored in these RAM chips are
maintained by battery power, and they
are generally not accessible to the operating
system.
COMPARE
TTL (Transistor-Transistor Logic)
M
CMOT (Common Management
Information Services and Protocol
Over TCP/IP)
An effort to implement the OSI framework's
CMIS and CMIP services on the Internet
MAJOR COMPONENTS IN THE ISO-OSI NETWORK MANAGEMENT MODEL


Prev Page 199 Next

CNI (Certified NetWare Instructor)
187
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
community's TCP/IP protocol suite, rather
than on OSI layer protocols. For various
reasons, including the popularity of SNMP
and the difficulty of porting the OSI model
to a TCP/IP environment, CMOT was never
completed.
M
CMS (Conversational Monitor System)
A subsystem in IBM's SNA.
SEE
SNA (Systems Network Architecture)
M
CN (Common Name)
In the NetWare Directory Services (NDS) for
Novell's NetWare 4.x, a name associated
with a leaf object in the NDS Directory tree.
For a user object, this would be the user's
login name.
M
CNA (Certified NetWare
Administrator)
A title given to people who successfully com-
plete Novell-authorized courses on adminis-
tering a NetWare network and/or pass a
comprehensive exam about this topic. The
CNA program is designed for people who
are responsible for the day-to-day opera-
tions and high-level maintenance of their
networks. CNAs must know how to add
and remove users, grant user rights, load
applications, do backups and other mainte-
nance tasks, and maintain network security.
Separate tests are required and degrees are
offered for NetWare 2.2, 3.11, and 4.x
environments.
CNAs are discussed in The CNA Study
Guide (James Chellis, et al. Network Press,
1996).
SEE ALSO
CNE; CNI; ECNE
M
CNE (Certified NetWare Engineer)
A title given to people who successfully com-
plete a whole series of Novell-authorized
courses on becoming technicians or consul-
tants for NetWare networks and/or pass a
comprehensive exam about this topic. The
CNE program is designed for people who
are responsible for designing and installing
NetWare networks, and also for the low-
level maintenance tasks such as diagnostics,
troubleshooting hardware or networking
software, and so forth. Separate tracks are
available for NetWare 2.2, 3.11, and 4.x. In
addition to demonstrating mastery of basic
and advanced topics related to NetWare,
successful CNE candidates must demon-
strate mastery of networking technology
and operating system concepts.
SEE ALSO
CNA; CNI; ECNE
M
CNI (Certified NetWare Instructor)
A title given to people who successfully com-
plete a comprehensive and rigorous training
program in order to teach Novell courses.
Candidates who are accepted for the CNI
program must demonstrate a proficiency
in their area of specialization by attending
each course they want to teach and passing
the course test at a more stringent level than
is required of ordinary (CNA or CNE)
students.
As a final requirement, candidates must
pass an IPE (instructor performance evalua-
tion). Among other things, candidates must


Prev Page 200 Next

188
CO (Central Office)
set up a classroom or lab, and then teach a
45- to 60-minute section of the course for
which the candidate wants to become an
instructor. Candidates do not know which
section they will be asked to teach until the
day before their evaluation.
SEE ALSO
CNA; CNE; ECNE
M
CO (Central Office)
The telephone switching station nearest to a
customer (residential or business). Custom-
ers are connected directly to a CO, which
connects them to other points in the tele-
communications hierarchy. The CO pro-
vides services such as switching, dial tone,
private lines, and centrex.
M
Coax Booster
A device that strengthens the signal in a
coaxial cable, thereby making it possible
to run a cable over greater distances.
MCOCF (Connection-Oriented
Convergence Function)
In the DQDB (Distributed Queue Dual-
Bus) network architecture, a function that
prepares data coming from or going to a
connection-oriented service. The service first
establishes a fixed, but temporary, connec-
tion, then transmits the data, and finally
breaks the connection.
MCodec
A codec is a device for converting analog
signals to digital form. For example, codecs
are used in digital telephone systems, such as
ISDN (Integrated Services Digital Network),
so that voice signals can be transmitted over
digital lines. The name is a contraction of
coder/decoder.
To make the conversion, a codec must
use some type of signal-sampling technique.
These samples are converted into discrete
signals for transmission across the digital
lines.
The most common conversion method
is PAM (pulse amplitude modulation), in
which samples of the analog signal's ampli-
tude are converted into discrete signals
whose amplitude corresponds to the analog
signal's amplitude at sampling time. To
reproduce the original signal accurately,
PAM devices must sample the analog signal
at a rate at least twice the frequency's signal.
For example, for voice signals, which have a
4 kilohertz bandwidth, the PAM device must
sample at least 8,000 times.
The discrete amplitude value is modu-
lated one more time to make it compatible
with the digital circuits. PCM (pulse code
modulation) converts the PAM signals into a
stream of binary values. To make this con-
version, the range of amplitudes in a PAM
signal is divided into 128 discrete quantizing
levels.
To represent 128 possible amplitude val-
ues, seven bits are needed for each PAM sig-
nal. This means that PCM must work at 56
kilobits per second (kbps) or faster. Digital
channels in North America provide a 64
kbps capacity, which means 8 kbps can be
used for administrative and system control
purposes.


Prev Page 201 Next

Collision Detection and Avoidance
189
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
BROADER CATEGOR Y
Digital Communication
SEE ALSO
Modulation
M
Code Excited Linear Predictive Coding
(CELP)
SEE
CELP (Code Excited Linear Predictive
Coding)
M
Coding
Coding is a general term for a representa-
tion, usually by means of a predefined syn-
tax or language. For example, in the OSI
Reference Model, an application layer
packet, or protocol data unit (APDU), will
have a coding that depends on the applica-
tion involved.
ASCII and EBCDIC are two widely used
codings. Abstract Syntax Notation One
(ASN.1) coding is used in many contexts
that adhere to the OSI Reference Model,
such as in network management tasks.
In a communications setting, several
types of coding are distinguished, and each
type may occur dozens of times:
Source: The coding used by the applica-
tion that initiates a transmission. That
application must be running on an end
system-that is, on a network node
capable of using all seven layers in the
OSI Reference Model.
Target: The coding used by the applica-
tion that receives a transmission. The
receiving application must be running
on an end system.
Transfer: A coding used by the applica-
tions at both ends of the connection or
by the translation program. Transfer
coding may be needed if the source
and target codings are different.
SEE ALSO
ASCII; ASN.1; EBCDIC
MCold Boot Loader
In Novell's NetWare, a program on the file
server's hard disk that will automatically
load NetWare after a cold boot.
M
Collision Detection and Avoidance
In an Ethernet network, a collision is the
simultaneous presence of signals from two
nodes on the network. A collision can occur
when two nodes each think the network
is idle and both start transmitting at the
same time. Both packets involved in a colli-
sion are broken into fragments and must be
retransmitted.
To detect a collision, nodes check the DC
voltage level on the line. A voltage level two
or more times as high as the expected level
indicates a collision, since this means there
are multiple signals traveling along the wires
at the same time. Collision detection in
broadband networks involves a separate
bandwidth for collision detection and is
somewhat more complex, since there may
not be any DC voltage to test.
In the CSMA/CD (carrier sense multiple
access/collision detection) media-access
method, for example, collision detection
involves monitoring the transmission line for
Collision Detection


Prev Page 202 Next

190
COM (Common Object Model)
special signals that indicate that two packets
were sent onto the network at the same time
and have collided. When this happens, spe-
cial actions are taken (as described in the
CSMA/CD article).
To avoid collisions, nodes can send special
signals that indicate a line is being used for a
transmission. For example, the CSMA/CD
media-access method uses RTS (Ready To
Send) and CTS (Clear To Send) signals
before sending a frame onto the network.
A node transmits only after the node has
requested access to the line and been granted
access. Other nodes will be aware of the
RTS/CTS transmission and will not try to
transmit at the same time.
BROADER CATEGORIES
CSMA/CD (Carrier Sense Multiple
Access/Collision Detect); Ethernet
MCOM (Common Object Model)
COM is an object-oriented, open architec-
ture that is intended to allow client/server
applications to communicate with each
other in a transparent manner, even if these
applications are running on different plat-
forms. Objects can also be distributed over
different platforms.
The COM model is a joint project of
Microsoft and Digital Equipment Corpora-
tion (DEC). Their immediate goal is to allow
networks or machines that use Microsoft's
Object Linking and Embedding (OLE)
technology to communicate transparently
with networks or machines that use DEC's
ObjectBroker technology.
To provide the cross-platform capabili-
ties, COM uses OLE COM, a protocol
based on the DCE/RPC (Distributed
Computing Environment/Remote Procedure
Call) protocol. Once implemented, COM
will allow machines running Microsoft
Windows, Windows NT, and Macintosh
environments to communicate in a trans-
parent manner with machines running
DEC's OpenVMS operating system or
any of several UNIX implementations.
COMPARE
ObjectBroker; OLE (Object Linking and
Embedding)
M
COM1, COM2, COMx
On a PC, the names associated with succes-
sive serial ports. Devices that might be
connected to such a port include modems,
pointer devices, and some printers.
COMPARE
LPT1
M
Combiner
A combiner is a fiber-optic coupler (optical
signal splitter and redirector) that combines
multiple incoming signals into a single out-
going signal.
A particular type of combiner is an essen-
tial element for WDM (wavelength division
multiplexing), in which signals from multi-
ple channels are sent over the same output
channel. The input channels are all trans-
mitting at different wavelengths, and the
coupler's job is to combine the signals in the
proper manner. A combiner is sometimes
known as a combiner coupler.
Collision Avoidance


Prev Page 203 Next

Communication, Asynchronous
191
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
SEE ALSO
Coupler
M
Command Line Utility (CLU)
SEE
CLU (Command Line Utility)
M
Commercial Internet Exchange (CIX)
SEE
CIX (Commercial Internet Exchange)
M
Committed Information Rate (CIR)
SEE
CIR (Committed Information Rate)
MCommon Carrier
A private company, such as a telephone
company, that supplies any of various com-
munications services (telephone, telegraph,
Teletex, and so on) to the public.
M
Common Mail Calls (CMC)
SEE
CMC (Common Mail Calls)
M
Common Name (CN)
SEE
CN (Common Name)
M
Common Programming Interface
for Communications (CPIC)
SEE
CPIC (Common Programming Interface
for Communications)
M
Common User Access (CUA)
SEE
CUA (Common User Access)
M
Communicating Application
Specification (CAS)
SEE
CAS (Communicating Application
Specification)
M
Communication, Asynchronous
Asynchronous communications are those in
which a transmission may take place at a
variable rate, and in which byte boundaries
are indicated by a combination of start and
stop bits. Transmission elements are distin-
guished by these special bits. This is in con-
trast to synchronous communication, in
which transmission elements are identified
by reference to a clock or other timing
mechanism.
Examples of asynchronous processes
include voice or data transmissions
(commonly using modems), terminal-host
communications, and file transfer. Modems,
terminals, pointer devices, and printers
are all devices that use asynchronous
communications.
In asynchronous communication, the
occurrence of the special start bit indicates
that a byte is about to be transmitted. The
duration of the start bit indicates the length
of a bit interval (duration of a single signal
value), which represents the speed at which
that byte is going to be transmitted. In a
sense, asynchronous transmissions synchro-
nize for each byte.


Prev Page 204 Next

192
Communication, Bisynchronous
With respect to the communication,
both sender and receiver need to agree
on the number of start and stop bits, and
also on whether a parity bit will be used.
This information is necessary to identify the
transmission elements. If a parity bit is used,
knowing what kind of parity is operating
will help interpret the transmission contents.
Asynchronous transmissions are less effi-
cient than synchronous (time-based) ones.
For example, the start and stop bit around
each byte represent 25 percent overhead for
an asynchronous byte. Because of this lesser
efficiency, asynchronous communications
cannot attain the bandwidths possible with
synchronous transmissions.
On the other hand, asynchronous trans-
missions are much more flexible, forgiving,
and easier to correct than the faster moving
synchronous transmissions.
SEE ALSO
Communication, Synchronous
MCommunication, Bisynchronous
In bisynchronous, or bisync, communica-
tion, a special (SYN) character is used to
establish synchronization for an entire data
block. Both sender and receiver must be syn-
chronized. The receiver must acknowledge
the receipt of each block with alternating
ACK characters: ACK0 for one block,
ACK1 for the next, ACK0 for the next, and
so on. Two successive acknowledgments
with the same ACK character indicate a
transmission error.
Also known as BSC, bisynchronous
communication is used in IBM mainframe
environments. It is used primarily when
transmitting data in EBCDIC format.
M
Communication Buffer
RAM set aside on a file server for tempo-
rarily holding packets until they can be
processed by the server or sent onto the net-
work. The RAM will be allocated as a num-
ber of buffers, each with a predetermined
size. A communication buffer is also known
as a routing buffer or packet receive buffer.
M
Communication Medium
The physical medium over which a commu-
nications signal travels. Currently, the most
popular medium is cable. Wireless media,
such as infrared wave, microwave, or radio
wave, are also becoming more widely used.
MCommunication, Synchronous
Synchronous communications are those that
depend on timing. In particular, synchro-
nous transmissions are those that proceed at
a constant rate, although this rate may
change during different parts of a communi-
cation (or when the line quality changes).
In synchronous communications, trans-
mission elements are identified by reference
to either an external clock or self-clocking,
signal-encoding scheme. This is in contrast
to asynchronous communication, in which
transmission elements are identified by spe-
cial signal values (start and stop bits).
Synchronous communications can
achieve very large bandwidths, eventually
allowing speeds of over 100 Mbps. Unfortu-
nately, as transmission rate increases, signal
quality decreases, because each bit interval
becomes extremely short.


Prev Page 205 Next

Computer
193
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
When an external clock is used for synchro-
nous communications, the duration of test
bits are timed, and the resulting values are
used as the bit-interval value. It is necessary
to resynchronize the transmission occasion-
ally to make sure that the parties involved
do not drift apart in their timing. This is a
real danger, because even tiny differences in
timing can have a significant effect when
millions of bits are transferred every second
in a communication.
To avoid such a problem, many synchro-
nous transmission methods insist that a sig-
nal must change at least once within a
predetermined amount of time or within
a given block size. For example, the B8ZS
(bipolar with 8 zero substitution) signal-
encoding scheme is based on a requirement
that a transmission can never contain more
than seven 0 bits in succession. Before that
eighth consecutive 0, a 1 bit will be inserted.
Self-clocking, signal-encoding schemes have
a transition, such as a change in voltage or
current, in the middle of each bit interval. A
self-clocking encoding method changes the
signal value within every bit interval to keep M
the two parties in synch during a transmis-
sion. This works because each party can
recalibrate its timing if it notices a drift.
Self-clocking methods avoid the need to
insert extra bits (as in the B8ZS encoding
scheme). On the other hand, a self-clocking
machine needs a clock at least twice as
fast as the transmission speed in order to
accomplish the signal changes within each
bit interval. Expressed differently, this
means you will not be able to transmit any
faster than at half the clock speed on a
machine. (You can effectively increase the
speed by compressing files before transmis-
sion, thereby sending more information than
the bit rate would indicate.)
COMPARE
Communication, Asynchronous
M
Compatibility
Compatibility is the ability of one device or
program to work with another. Compatibil-
ity is sometimes built into the product; in
other cases, the compatibility is achieved
through the use of drivers or filters.
For example, to ensure that a network
interface card will work with a network
software package, drivers are used. Rather
than creating drivers for every adapter, a
more common strategy is to create a more or
less generic driver interface, and then try to
get developers to adapt the interface for
their products to this generic interface. Ven-
dors may also adapt the generic drivers to
handle the special features of particular
products.
CompuServe
SEE
CIS (CompuServe Information Services)
MComputer
Networks consist of computers, along
with some means for connecting the com-
puters and enabling them to communicate
External Clocks
Self-Clocking Transmissions


Prev Page 206 Next

194
Computer
with each other. The figure "Context of
computers in networks" shows the role
of computers.
The individual computers that make up
a network are known as nodes, or stations.
Nodes can be PCs, minicomputers, or even
mainframes.
The term PC can refer to any type of per-
sonal computer, but there are differences
between, for example, a network using IBM
PC and compatible machines and one using
Macintoshes. Both of these networks will, in
turn, differ somewhat from a network that
uses Sun workstations.
Almost all PC-class machines are based
on one of three processor families: the Intel
80x86 family (including the analogous pro-
cessors from third-party manufacturers and
Pentium, the newest incarnation from Intel),
the Motorola 680x0 family (used in the
Macintosh and some higher end work-
stations), and RISC (reduced instruction set
computing) chips (used in special-purpose
machines, number crunchers, and high-end
workstations but starting to migrate down
to lower-level machines).
Unless otherwise stated, PC will refer to
the IBM PC and compatible computers (as
well as to IBM's own Micro Channel Archi-
tecture line of computers) based on the Intel
architecture. Where the discussion concerns
Macintoshes or Sun machines, this will be
mentioned.
PCs can be servers, workstations, or inter-
network links in a network. The whole
gamut of PCs can be used in a network. You
can even attach a palmtop computer to a
network. Not all PCs can serve all functions
in a network, however.
To work in a network, PCs need a special
network interface card (NIC), or adapter.
This component provides the appropriate
chips and circuitry for translating com-
mands or data into packets and then into
electrical signals to be sent over the network.
At the receiving end, the NIC captures the
received transmission, and again translates,
but this time from the electrical format used
on the network to a format the networking
software understands.
A node may function as a workstation,
a server, or an internetwork link (which
serves to connect two or more networks). In
certain combinations, a computer can serve
more than one of these functions at a time.
CONTEXT OF COMPUTERS
IN NETWORKS
PCs


Prev Page 207 Next

Computer
195
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
A server provides access to resources or ser-
vices, such as files, printers, fax machines,
electronic mail, and so on. Servers may be
distinguished by the elements to which they
control access. For example, you will see
references to file servers, print servers, fax
servers, and communications servers. A file
server generally runs the network, providing
access to programs and data, and sometimes
also to peripherals.
A network need not have a server. If each
node is a workstation, then each node is
accessible to other nodes. Networks in
which all nodes are workstations are known
as distributed, peer-to-peer, or simply peer
networks. Artisoft's LANtastic, Novell's
NetWare Lite, and Microsoft's Windows for
Workgroups are examples of peer-to-peer
network packages.
If there is a server, it may be dedicated
or nondedicated. A dedicated server cannot
be used as a workstation. Networks with a
dedicated server are known as centralized
networks or server-based networks.
A workstation requests access to files, print-
ers, and so on, from a server. Actually, the
user simply requests such services as if they
were available on the workstation itself.
Special shell and redirection software will
route the request to the server. Users can
also use a workstation for non-network
activity.
There is no inherent hardware difference
between a server and a workstation. Practi-
cal performance considerations, however,
dictate that servers should be faster, more
Server
Workstation
powerful machines. In practice, worksta-
tions may be any level PC, with 80286 and
80386 being the most common. Servers are
almost always 80386 or 80486 machines.
In fact, some network operating systems
require at least an 80386 processor for the
server.
A special class of machines, called disk-
less workstations, can be used only as work-
stations on a network. These workstations
have their boot instructions in ROM, boot
to the network, and can be used only to do
work on the network. Since they do not
have disk drives, you cannot download any
data to the workstations or upload data to
the network.
An internetwork link serves to connect two
networks to each other. A PC may serve as
an internetwork link and as a server or
workstation at the same time. Examples of
internetwork links include bridges, routers,
brouters, and gateways.
Networks can include minicomputers, such
as the DEC VAX or the IBM AS/400 series,
or mainframes, such as the IBM System/370
and System/390 families (although this is
more common in older networks and in net-
works run by or from MIS departments).
Many networks, particularly those in
large organizations, include minicomputers
and mainframes. For example, it is not
uncommon to see a minicomputer serving
as a front-end processor (FEP) for a main-
frame, to handle incoming transmissions
from PCs or terminals.
Internetwork Link
Non-PC


Prev Page 208 Next

196
Computer-Based Message System (CBMS)
Mainframe- and PC-based networks
are very different worlds from each other,
with different character codes, protocols,
frame formats, and operating environments.
Despite (or perhaps because of) the obstacles
that have always existed to make PC-
mainframe communication such a challenge,
there are a frightening number of possible
configurations in which a PC can talk to a
mainframe or minicomputer. IBM alone has
dozens of hardware and software products
(such as the SNA architecture and the IBM
Data Connector) for helping computers of
various sizes communicate with each other.
BROADER CATEGOR Y
Hardware Network
M
Computer-Based Message System
(CBMS)
An older term for a message handling system
or for electronic mail.
M
Computer Business Manufacturers
Association (CBEMA)
SEE
CBEMA (Computer Business
Manufacturers Association)
M
Computer-to-PBX Interface (CPI)
SEE
CPI (Computer-to-PBX Interface)
MConcentrator
Most generally, in the area of communica-
tions, a concentrator is a device that can
take multiple input channels and send their
contents to fewer output channels. In addi-
tion to these multiplexing capabilities, a
concentrator can store data until an output
channel becomes available.
In networking hardware, a concentra-
tor is essentially an upscale hub. The terms
hub and concentrator are often used inter-
changeably, and the term wiring center is
often used to refer to either a hub or a
concentrator.
As is the case for a hub, the main function
of a concentrator is to serve as a termination
point for cable running from individual
nodes (stations) in a network. The cable
connects to the network or to another wir-
ing center.
A concentrator may have multiple boards
or boxes mounted on a rack. Each board is
essentially a hub-a wiring center for a sin-
gle network's nodes. Such boards generally
include LEDs (light-emitting diodes) to indi-
cate the status of each port on the board.
The size and complexity of the concentra-
tor depends on the number of boards that
have been installed. Partly because of their
versatility and power, high-end concentra-
tors can cost as much as $50,000.
Hubs and concentrators can be viewed as
the ends of a continuum. Hub manufactur-
ers are likely to include concentrators in
their product lines.
Concentrators can be much more versatile
than hubs in what they can connect. For
example, a concentrator might connect net-
work elements (or networks) with different
cabling and perhaps even with different
architectures.
Concentrator Operation


Prev Page 209 Next

Configuration Management
197
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
Note that the concentrator might not
necessarily be connecting these different
architectures to each other. Rather, the con-
centrator may be serving as a wiring conduit
for multiple (independent) networks simul-
taneously; for example, for networks run-
ning in different departments in a company.
It is possible to include bridging or routing
capabilities in the concentrator. With
bridging or routing, a concentrator can con-
nect different architectures to each other.
Concentrators are generally located in
a wiring closet, which serves as a wire-
collection location for a predefined area.
In the closet, the concentrator may be con-
nected to another concentrator, to an inter-
mediate distribution frame (IDF), to a main
distribution frame (MDF), or perhaps to a
telephone line. IDFs collect the wiring from
a limited area (such as a floor) and feed this
to the MDF for the building. The MDF con-
nects the building to the outside electrical
world.
All concentrators provide connectivity, serv-
ing as wiring centers. Many concentrators
also have their own processor and can serve
as network activity monitors. Concentrators
with processors save performance and other
data in a management information base
(MIB). This information can be used by
network management software to fine-tune
the network.
A board in the concentrator may have its
own processor for doing its work. In such
a case, the board is using the concentrator
as a convenient location to use as a base
of operations.
Concentrator Features
BROADER CATEGOR Y
Intranetwork Link
SEE ALSO
Hub; Wiring Center
M
Conductor
Any material (for example, copper wire)
that can carry electrical current. Compare
conductor with semiconductor or insulator.
SEE ALSO
Cable
M
CONFIG.SYS
In DOS and OS/2 environments, CON-
FIG.SYS is a file that contains information
about various types of configuration and
driver settings. For example, CONFIG.SYS
may include information about drivers and
memory managers that are loaded into
memory. The OS/2 configuration file can be
quite long and complex.
MConfiguration Management
Configuration management is one of five
OSI network management domains specified
by the ISO and CCITT. Configuration man-
agement is concerned with the following:
I Determining and identifying the
objects on the network and their
attributes
I Determining states, settings, and other
information about these objects
I Storing this information for later
retrieval or modification


Prev Page 210 Next

198
Configuration Management
I Reporting this information if requested
by an appropriate and authorized pro-
cess or user
I Modifying the settings for objects,
if necessary
I Topology management, which involves
managing the connections and rela-
tionships among the objects
I Starting up and shutting down net-
work operations
The first task for configuration manage-
ment is to identify objects such as stations,
bridges, routers, and even circuits. Depend-
ing on the sophistication of the management
package, this process may be automatic or it
may be done manually.
Each object will have configuration states
and other information associated with it.
For example, a node might have the follow-
ing settings:
I Interface settings, such as speed, parity,
jumper settings, and so on
I Model and vendor information,
including serial number, operating
system, memory and storage, hard-
ware address, and so on
I Miscellaneous details, such as installed
drivers and peripherals, maintenance
and testing schedules, and so on
Similarly, leased lines or circuits will have
information such as identification number,
vendor (or leaser), speeds, and so on.
Identifying Objects
and Determining Settings
Within the OSI model, four operational
states are defined for an object:
Active: The object is available and in use,
but has the capacity to accept services
or requests.
Busy: The object is available and in use,
but currently is not able to deal with
any more requests.
Disabled: The object is not available.
Enabled: The object is operational and
available, but not currently in use.
Such values must be determined-manu-
ally or automatically-and stored for easy
access and updating (for example, in a rela-
tional database). If stored in a database,
the information will generally be accessed
using some type of query language. SQL
(Structured Query Language) has become a
standard means for accessing object opera-
tional state information. The configuration
management capabilities include being able
to report this information upon request.
The values and states associated with net-
work objects may be changed. For example,
they will be changed when trying to commu-
nicate with a network object, correct a fault,
or improve performance. Certain values (for
example, state information) may be changed
automatically when an action is begun on
the network. Other values may need to be
changed by the system administrator.
BROADER CATEGOR Y
Network Management
Operational States for an Object
Modifying Settings and States


Prev Page 211 Next

Connectionless Service
199
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
SEE ALSO
Accounting Management; Fault Manage-
ment; Performance Management;
Security Management
MConfiguration, Network
Network configuration consists of the equip- M
ment, connections, and settings in effect for
a network at a particular time. Equipment
generally refers to hardware (computers,
peripherals, boards, cables, and connectors),
but may also include software under certain
circumstances.
Because compatibility and interoperabil-
ity can sometimes be elusive in the network-
ing world, a system administrator needs
to know considerable detail about the
equipment on the network. This informa-
tion may include specific model numbers,
memory specifications, enhancements, and
so on. This information must be updated
scrupulously or conflicts may occur. Fortu-
nately, most networking systems include a
utility for recording configuration informa-
tion and for updating it as the network
changes.
The current settings for each piece of
equipment should also be recorded as part
of the configuration information. When
deciding on specific settings, it is important
to avoid conflicts. A conflict can arise, for
example, because two boards each wanted
to use the same memory location or inter-
rupt line. Again, most network operating
systems include a utility to help keep this
information organized and to spot potential
conflicts before they are made official.
M
Conformance Requirements
The set of requirements a device or imple-
mentation must satisfy in order to be
regarded as conforming to a particular
specification or recommendation.
Congestion
In data communications, a state in which the
data traffic approaches or exceeds the chan-
nel's capacity, resulting in a severe perfor-
mance degradation and, possibly, loss of
packets.
MConnectionless Service
In network operations, a connectionless ser-
vice is one in which transmissions take place
without a preestablished path between the
source and destination. This means that
packets may take different routes between
the source and destination. Connectionless
services are defined at the network and
transport layers, with the specifications in
CLNS (Connectionless Mode Network Ser-
vice) and CLTS (Connectionless Transport
Service), respectively.
Because packets may arrive by different
paths and in random sequences, there is no
way to guarantee delivery in connectionless
service. Instead, the higher layers, particu-
larly the transport layer, are left with the job
of making sure packets reach their destina-
tion without error.
CLNP (Connectionless Network Pro-
tocol), CLTP (Connectionless Transport
Protocol), IP (Internet Protocol), and UDP
(User Datagram Protocol) are examples


Prev Page 212 Next

200
Connectionless Transport Service (CLTS)
of protocols that support connectionless
service.
COMPARE
Connection-Oriented Service
M
Connectionless Transport Service
(CLTS)
SEE
CLTS (Connectionless Transport Service)
MConnection, Network
A network connection is a linkage between
network elements. Network connections
exist on two different levels:
Physical connections: Concern the cables
and connectors (used to create the
physical topology of the network)
and the machines connected. When
building a network, you must first
establish the physical connections.
Logical connections: Concern the way
in which nodes on the network
communicate with each other. For
example, the sequence in which a
token is passed in an ARCnet or Token
Ring network depends on the net-
work's logical topology, not on the net-
work's physical layout. Thus, node x
may communicate with node y in the
network, even though the two nodes
are not adjacent machines in the physi-
cal network.
M
Connection Number
A number assigned to any node that attaches
to a file server. The network operating
system on the file server uses the connection
number to control how nodes communicate
with each other. A node will not necessarily
be assigned the same connection number
each time it attaches to the network.
MConnection-Oriented Service
In network operations, a connection-
oriented service is one in which a connec-
tion (a path) must be established between
the source and destination before any data
transmission takes place. With this service,
packets will reach their destination in the
order sent, because all packets travel along
the same, "no-passing" path.
With this type of connection, the OSI
data-link layer, for example, checks for
errors, does flow control, and requires
acknowledgment of packet delivery.
X.25 and TCP (Transmission Control
Protocol) are two protocols that support
connection-oriented services. Connection-
oriented services are defined at the network
(CONS) and transport (COTS) layers.
COMPARE
Connectionless Service
M
Connectivity
The ability to make hardware and/or
software work together as needed. The
principles and details of how this happens
comprise about half of this book and thou-
sands of pages in other books.


Prev Page 213 Next

Connector
201
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
M
Connector
A connector provides the physical link
between two components. For example, a
connector can link a cable and a network
interface card (NIC), a cable and a trans-
ceiver, or two cable segments. For electrical
cable, a connection is established whenever
the conducting wires (or extensions) from
the two connectors make and maintain
contact. The signal can simply move across
the contact.
For fiber-optic cable, good connections
take much more work, because the degree of
fit between the two fiber cores determines
the quality of the connection. This fit cannot
be taken for granted, because the diameters
involved are smaller than a human hair.
Connectors differ in their shape, size, gen-
der, connection mechanism, and function.
These features influence, and sometimes
determine, where a connector can be used.
Where necessary, special adapters may be
used for connections involving different con-
nector combinations. For example, N-series
to BNC adapters make it possible to connect
thick to thin coaxial cable.
Connectors also differ in how sturdy they
are, how easily and how often they can be
attached and detached (how many matings
they can survive), and in how much signal
loss there is at the connection point.
The type of connector needed in a partic-
ular situation depends on the components
involved and, for networks, on the type
of cable and architecture being used. For
example, an Ethernet network using co-
axial cable will need different connectors
between cable and NIC than an IBM Token
Ring network using shielded twisted-pair
(STP) cable.
The world of connectors includes its own
miniworld of acronyms: N, BNC, DB, DIN,
RJ, SC, SMA, ST, TNC, V.32, and so on. To
make matters even more confusing, some
connectors have more than one name.
About half a dozen types of connec-
tors are used with electrical cable in some
network-related contexts; about a dozen
more types are used with fiber-optic cable.
These connector types are discussed in sepa-
rate articles. This article discusses connec-
tors in general.
A connector may be passing the signal
along or absorbing it (as a terminator does).
A connector that passes a signal along
may pass it unmodified or may clean
and boost it.
Connectors can serve a variety of pur-
poses, including the following:
I Connect equal components, such as
two segments of thin coaxial cable
I Connect almost equal components,
such as thin to thick coaxial cable
I Connect unequal components, such
as coaxial to twisted-pair cable
I Connect complementary components,
such as an NIC to a network
I Terminate a segment; that is, connect
a segment to nothing
I Ground a segment; that is, connect a
segment to a ground
Connector Functions


Prev Page 214 Next

202
Connector
In this context, the term shape refers to
the component, not to the connection. Spe-
cially shaped connectors are used for partic-
ular types of connections or for connections
in particular locations. For example, a
T-connector attaches a device to a cable
segment; an elbow connector allows wiring
to meet in a corner or at a wall.
The connector shapes used in network-
ing setups are listed in the table "Cable
Connector Shapes," and the figure "Some
connector shapes" shows examples.
Connector Shapes
Connector gender basically refers to whether
a connector has plugs or sockets. The gender
is important because the elements being con-
nected must have complementary genders.
A male connector is known as a plug; the
female connector is known as a jack. With
a few notable exceptions, such as the IBM
data connectors and certain fiber-optic
connectors, all connector types have dis-
tinct genders. The figure "Connector gen-
ders" shows examples of male and female
connectors.
Connector Genders
CABLE CONNECTOR SHAPES
SHAPE
DESCRIPTION
Barrel
DB- or
D-type
Elbow
RJ
T
Y
Miscellaneous
Used to link two segments of cable in a straight run; i.e., in a location where there are no cor-
ners or turns. In networking, BNC and N-series barrel connectors are used to connect sec-
tions of thin and thick coaxial cable, respectively.
Describes the connector's frame and refers to a whole family of connectors most commonly
used for serial, parallel, and video interfaces. DB-9 and DB-25 connectors are used for serial
ports on ATs and XTs. 9-pin versions are used for connecting a monitor to the video board.
External network cards, which attach to the parallel port, use DB connectors.
Connector with a right-angle bend, used to connect two sections of cable in a corner or to
accomplish a change of direction.
Used to connect telephones to the wall or to modems. RJ-11 and RJ-45 are two commonly
used types.
Used to attach a device to a section of cable. The horizontal bar of the T links two sections of
cable, like a barrel connector; the vertical bar attaches the device. In networks, a T-connector
is used to link a section of drop cable to the main cable segment in a thick Ethernet network.
Sometimes used in multiplexers; for example, in a component that provides two ports from
one. The shape is mainly a matter of convenience.
There are no inherent limitations in the shape a connector can have. Special-shaped connec-
tors can be used when necessary.


Prev Page 215 Next

Connector
203
a
b
c
d
efghijklmno
p
q
rstuvwxyz
SOME CONNECTOR SHAPES
CONNECTOR GENDERS


Prev Page 216 Next

204
Connector
The connection mechanism defines how the
physical contact is made to allow the signal
to pass from one side of the connection to
the other.
Connection mechanisms differ consider-
ably in how sturdy they are. For example,
the pin-and-socket connection at a serial
port can be wobbly without extra support
from screws. On the other hand, fiber-optic
Connection Mechanisms
connectors must be cut to precise propor-
tions, and must not allow any play in the
connection, since a cable thinner than a
human hair does not need much room to
move around.
Connectors are not necessarily named
according to the connection mechanism.
Rather, the names may have some other
basis. The table "Selected Connector
Types" illustrates the range of connection
mechanisms.
COMPONENTS FOR OTHER TYPES OF LINKS
Connectors connect equal or complementary components. The following components make other types of
links possible:
Cable Adapters: Connect almost equal components. Adapters mainly serve to allow size adjustments.
Terminators: Absorb a signal at the end of a network or cable segment to prevent the signal from being
reflected back into the cable (thereby causing interference with newer signals traveling out on the cable).
Networks have stringent rules about what must be terminated; it's very wise to observe these rules.
Grounded Terminators: Work just like regular terminators, except that grounded terminators have a pig-
tail or a small metal chain at the end. This needs to be attached to a suitable object to dissipate the charge and
to prevent it from being stored up anywhere. (One end of any network or segment must be grounded as well
as terminated.)
Baluns: Connect unequal components; that is, components that have different electrical properties (imped-
ances). Baluns are commonly used to connect coaxial to twisted-pair cable.
Transceivers: Connect components and also process signals. Transceivers are receivers and transmitters.
Because their main function is passing information (rather than connecting), transceivers may be installed
directly on the network interface card. Transceivers establish an electrical, rather than merely a physical,
connection.
Repeaters: Clean and boost a signal before passing the signal on to the next cable segment or node. There
are often limitations on how repeaters may be distributed on a network. For example, the IEEE 802.3 stan-
dards allow at most four repeaters on the signal path between any two stations on an Ethernet/802.3 net-
work. Repeaters are primarily signal boosters, and are connectors only secondarily. Like transcievers,
repeaters establish an electrical connection.


Prev Page 217 Next

Connector
205
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
SELECTED CONNECTOR TYPES
TYPE
DESCRIPTION
BNC (bayonet
nut connector)
TNC (threaded
nut connector)
N-series
Centronics
D-type
V.35 and M.50
DIN
RJ-xx
IBM Data
Slide together and then lock into place. Ethernet networks with thin coaxial cable use
BNC connectors. A variant on the standard BNC connector is used for twinaxial
cable. BNC connectors can survive many matings.
Similar to BNC in construction, except that TNC has threads instead of notches, which
create tighter connections.
Similar to TNC, except that the barrel is somewhat fatter and the plug is somewhat thinner.
N-series connectors are used with thick coaxial cable in thick Ethernet networks. N-series
connections are quite tight.
Use teeth that snap into place. The printer end of a parallel PC-printer connection usually
has this type of connector. IEEE-488 interfaces also use Centronics connectors. The term
Telco-type is also used to describe certain Centronics connectors.
One of the three classes of connectors that use pins and sockets to establish contact
between the elements involved. These are so named because the frame around the pins
and sockets that make up the connection resembles a D. The connectors for the serial and
parallel ports on most PCs use D connectors.
Also use pins and sockets, but they are arranged somewhat differently than for the D-type
connectors. V.35 connectors have more rectangular frames.
Round, but also use pins and sockets. The keyboard connector on most PCs is a DIN
connector, as are two of the connectors used for LocalTalk networks.
Connect by catching and locking a plug in place with an overhanging element in the jack
connector. RJ-xx, or modular, connectors are used in telephone connections and also with
twisted-pair cable in networks. Connector versions differ in the number of line pairs they
support, e.g., RJ-11 connectors support two pairs; RJ-45 connectors support up to four
pairs. A variant on this type is the MMJ (for modified modular jack) connector, which is
used in some DEC networks.
A specially designed connector used in IBM Token Ring networks. The connector has a
somewhat intricate connection mechanism that can short-circuit when disconnected, so
that the network can preserve its structure even when nodes drop out.


Prev Page 218 Next

206
Connector, AUI (Attachment Unit Interface)
These connection classes are all used
for electrical cable. Several of the same con-
nection principles also apply to fiber-optic
cable. Numerous types of fiber-optic connec-
tors exist, as discussed in the Connector,
Fiber-Optic article.
Attaching two connectors to each other
is known as mating. Because they involve
physical parts and are subject to wear and
tear, connectors become less effective as they
go through more matings. Because this can
lead to increased signal degradation, your
choice of connectors may depend on how
often you expect to connect and disconnect
network segments.
Another factor to consider is insertion
loss. The signal will undergo a certain
amount of loss and distortion at a con-
nection point. This insertion loss will be
expressed in decibels (dB). For electrical
connections, this value can be 15 dB and
more; for fiber-optic cable, this value will
generally be less than 1 dB.
SEE ALSO
Connector, AUI; Connector, BNC;
Connector, Fiber-Optic
MConnector, AUI (Attachment
Unit Interface)
An AUI connector is a 15-pin, D-type con-
nector that is used in some Ethernet connec-
tions. Typically, it is used to connect a drop
cable to a network interface card (NIC).
This type of connector is also known as a
DIX (for Digital, Intel, Xerox) connector.
The figure "An AUI connector" shows an
example.
The connection mechanism is the
D-type pin and socket, just as for the RS-
232 connectors found on most computers.
In addition, an AUI connector includes
a (sometimes fragile) slide mechanism
that can lock the connection into place.
MConnector, Barrel
A connector used to link two pieces of iden-
tical cable, such as thin or thick coaxial
cable. The name comes from the connector's
shape. BNC barrel connectors link thin
coaxial cable; N-series connectors link thick
coaxial.
Connector Mating and Insertion Loss
AN AUI CONNECTOR


Prev Page 219 Next

Connector, D-4
207
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
M
Connector, BNC
A BNC connector is used with coaxial
cable in thin Ethernet networks, in some
ARCnet networks, and for some video mon-
itors. Its name may come from Bayonet-
Neill-Concelnan, for its developers; from
bayonet nut connector, for its attachment
mechanism; or from bayonet navy connec-
tor, for one of its early uses. The figure "A
BNC connector" shows an example of this
type of connector.
To connect a BNC connector, you insert
the plug in the jack, and then lock in the
connection by turning the connector. The
simple plugging mechanism can survive
many matings, and the lock makes the
connection more stable.
BNC connectors come in the following
shapes and versions:
Barrel connector: Connects two pieces of
thin coaxial cable. Each end of the bar-
rel connector is typically female, which
means the cable pieces must have a
male BNC connector at the end being
attached.
Elbow connector: A BNC connector with
a right angle in it, for use in corners or
in other locations where the cabling
needs to change direction.
T-connector: Connects a network node
to the cable segment. The T-connector
usually has female connections at each
end and a male BNC connection form-
ing the descender in the T. A network
machine is attached to the male con-
nector; the other two ends are con-
nected to the trunk cable segment for
the network.
Terminator: Prevents a signal from
bouncing back from the end of the net-
work cable and interfering with other
signals. The terminator connects to a
BNC connector at the end of the trunk
cable segment.
Grounded terminator: Grounds and
terminates a thin Ethernet trunk seg-
ment. A grounded terminator connects
to a BNC connector at the end of a
trunk cable segment, but includes
a ground cable at the end of the ter-
minator. One end of each trunk cable
segment must be grounded.
M
Connector, D-4
A fiber-optic connector that uses a threaded
coupling nut for the connection.
SEE
Connector, Fiber-Optic
A BNC CONNECTOR


Prev Page 220 Next

208
Connector, D-type
MConnector, D-type
The D-type category of connectors is one of
the three classes of connectors that use pins
and sockets to establish contact between the
elements involved. These are so named
because the frame around the pins and sock-
ets that make up the connection resembles a
D. The connectors for the serial and parallel
ports on most PCs use D-type connectors.
D-type connectors are distinguished by
the number and arrangement of pins (and/or
sockets, depending on the connector's gen-
der) and by the size of the frame. Names
such as DB-9, DB-25, or DB-37 refer to con-
nectors with 9, 25, and 37 pins/sockets,
respectively.
Common types of D-type connectors
include the following:
I DB-9, which is used for some serial
(RS-232) interfaces and also for video
interfaces. The pin assignments are
different for these two uses, so the
connectors are not interchangeable.
I DB-15, which is used for video
interfaces.
I DB-25, which is used for some serial
(RS-232) interfaces and also for a par-
allel printer interface.
I DB-37, which is used for an RS-422
interface.
The figure "Examples of D-type connec-
tors" illustrates some of these types of con-
nectors. The actual pin assignments depend
on the cable's use.
In general, connections involving such
connectors can be flimsy unless the connec-
tors are locked into place with screws.
Special-purpose variants on the pin-and-
socket mechanism (and the D frame) have
special names. DIX (for Digital, Intel, and
Xerox), or AUI (for attachment unit inter-
face) connectors, are used in Ethernet net-
works. DIX connectors may also have a
slide mechanism to help lock the connection
into place.
M
Connector, Elbow
A connector with a right angle in it,
designed for connecting wires in a corner or
wherever a change of direction is needed.
M
Connector, ESCON (Enterprise
System Connection Architecture)
A fiber-optic connector for use with multi-
mode fiber in IBM's ESCON channel.
SEE
Connector, Fiber-Optic
EXAMPLES OF
D-TYPE CONNECTORS


Prev Page 221 Next

Connector, Fiber-Optic
209
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
M
Connector, F
A connector used in 10Broad36 (broad-
band Ethernet) networks and also in the
broadband versions of the (IEEE 802.4)
token-bus architecture.
M
Connector, FC
A connector used for fiber-optic cable,
which uses a threaded coupling nut for the
attachment and 2.5 millimeter ceramic fer-
rules to hold the fiber.
SEE
Connector, Fiber-Optic
M
Connector, Fiber-Optic
A fiber-optic connector must establish a
physical link between two segments of opti-
cal core, which are just a few nanometers
(billionths of a meter, or fractions of a
human hair) in diameter. The degree of over-
lap between the core segments determines
the quality of the connection, because this
overlap controls how much light is lost or
distorted in the crossover from one fiber to
the other. The figure "A fiber-optic connec-
tor" shows an example of this type of
connector.
A fiber-optic connection must not only
be precise and smooth, it must also be as
immobile as possible. Even the slightest
movement can cause unacceptable signal
loss. Fiber-optic connections should be put
through as few matings as possible, because
even a snug connection becomes less snug
each time it is made and unmade. (A mating
is the joining of two connectors.)

In fact, to encourage lifelong attach-
ments (instead of random matings), splices
are frequently used to make fiber-optic
connections. (A splice is a permanent con-
nection between two fiber segments.)
To establish a temporary but sound fiber-
optic connection, the following tasks are
necessary:
I Immobilize each fiber as completely as
possible.
I Polish the section that will make con-
tact to as smooth a finish as possible.
I Bring the fiber segments into maxi-
mum contact.
I Immobilize the connection.
An effective connector is one that has very
low insertion loss (signal loss that occurs as
the signal passes through the connector) and
very low return loss (signal that is reflected
back through the fiber from which the signal
came). Insertion losses of less than 1 decibel
A FIBER-OPTIC CONNECTOR
Features of an Effective
Fiber-Optic Connector


Prev Page 222 Next

210
Connector, Fiber-Optic
(dB), and usually less than 0.5 dB, are the
rule with fiber-optic connectors. This means
that almost 80 percent of the signal (almost
90 perecent with a 0.5 dB loss) gets past the
connector. In contrast, more than 90 percent
of an electrical signal may be lost going
through a connector.
The reflection loss indicates the amount
of the signal that is reflected back; that is,
the amount lost to reflection. A large nega-
tive decibel value means there was little loss
to reflection. For example, a reflection loss
of -40 dB means that 0.01 percent of the
signal was reflected back. By convention, the Polishing
negative sign is dropped when speaking of
loss; the -40 dB value is simply 40 dB. In
this case, and in several others involving
signals, a large positive decibel value is bet-
ter, even though the discussion involves loss.
Several components and steps are impor-
tant for making a satisfactory fiber-optic
connection. Ferrules help guide and immobi-
lize the fiber. To make a good connection,
the fiber ends must be properly and evenly
polished.
A ferrule grabs the fiber and channels it to
a point where it can be put in contact with
another fiber. The ferrule (which is derived
from a word for bracelet) is a thin tube into
which a segment of fiber is inserted. The
fiber will be trimmed and polished at the end
of the ferrule.
The best (and most expensive) ferrules are
made of ceramic. Ceramic is remarkably sta-
ble and well-behaved over the temperature
range the connector is likely to encounter
under ordinary conditions. Plastic is a
poorer (and cheaper) material for ferrules.
Stainless steel fits between these two
extremes in performance and price.
Even if the ferrule is designed to fit
as snugly as possible around the fiber,
there may still be movement because of
changes in temperature and humidity in
the area around the cable. To minimize the
movement produced by such climatic condi-
tions, the fiber may be glued to the ferrule
using epoxy, or wedged in more snugly by
slightly crimping the ends of the ferrule.
The fiber will be cut at the end of the ferrule.
On the fiber's scale, such a cut will look very
jagged and rough-unacceptable for making
a connection. To smooth the cut, the end
must be carefully and thoroughly polished.
Trying to polish the fiber ends to a com-
pletely flat surface is not always the best way
to make a clean connection. It is virtually
impossible to get both fiber ends smooth
enough and angled in the same direction. In
practice, there will always be gaps between
two smooth and flat surfaces.
A gap between the fiber ends will not only
result in a loss of the signal traveling on, it
will also cause more of the original signal to
be reflected back along the fiber. The return
reflection signal will interfere with the newer
signals moving along the fiber. Return reflec-
tion loss is one of the values that should be
as high as possible. As more of a reflected
signal is lost, less can actually be reflected
back. Losses of 30 to 40 dB are considered
good for this variable.
A relatively effective polishing strategy
aims for PC (physical contact) connections.
Ferrules


Prev Page 223 Next

Connector, Fiber-Optic
211
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
In this strategy, the ends of the fibers are pol-
ished to rounded ends. Such fibers will be in
physical contact, so there will be no air gap
to weaken the outgoing signal and reflect it
back.
Polishing can be a delicate and tedious
process, and is best left to the experts and
the machines.
Like electrical cable connectors, different
types of fiber-optic connectors have different
kinds of attachment mechanisms. The actual
attachments between ferrule shells may be
made by threading, snapping, or clicking.
In addition to attachment mechanisms,
fiber-optic connectors differ in the following
ways:
I The size of the ferrule.
I Whether the connector can be keyed.
Keying is a technique for making a
BUILDING YOUR OWN FIBER-
OPTIC CABLE CONNECTORS
If you'll be building fiber-optic cable connectors
yourself, keep in mind that both the epoxy glue
and crimping methods require considerable
skill and patience. Newer tools make the job
somewhat easier, but you still need to make sure
that the fiber is at exactly the right orientation
before gluing or crimping.
The fiber protruding through the tube needs to
be trimmed and polished so that the surface that
connects to the fiber in the other connector will
be as smooth as possible. The smoother the sur-
face, the better the connection you can make.
Types of Fiber-Optic Connectors
connector asymmetrical, usually by
adding a notch or plug. The asymme-
try makes it impossible to plug the
connector in incorrectly. It also ensures
that the fibers in the connector ends
always meet at the same orientation.
I
The number of matings the connectors
can endure without producing unac-
ceptable signal loss.
I Whether the fiber must be twisted
to make the connection. If it needs to
be turned, multiple fibers cannot run
through the same connector. Nontwist-
ing connectors are becoming much
more popular.
Connectors also differ in the way the fiber
is attached to the connector itself. You can
either use epoxy to glue the fiber into the
connector (usually into a tube, or ferrule), or
you can crimp the connector and the ferrule
together using a special tool. In general, fiber
that is attached to the connector using epoxy
glue is more robust and less likely to be
damaged than fiber attached by crimping.
Fiber-optic connectors can be a source of
significant signal loss, so it is important to
select connectors carefully. Find out how
many matings a fiber connector is specified
for. You should also make sure that the
cables you are connecting are as similar as
possible.
The table "Factors Contributing to Signal
Loss at Fiber-Optic Connectors" summa-
rizes problems that can arise with fiber-optic
connections. The sum of all these losses is
known as insertion loss and can be mea-
sured simply by taking readings of signal
strength at either end of the connection.


Prev Page 224 Next

212
Connector, Fiber-Optic
FACTORS CONTRIBUTING TO
SIGNAL LOSS AT FIBER-OPTIC CONNECTORS
FACTOR
DESCRIPTION
Core diameter
Core concentricity
Core ovality
NA mismatch
Lateral placement
Fiber cuts
Connection angle
Rough surface
Gaps
Contaminants
Bends
Promiscuity
Connecting a core with a given diameter to a core with a smaller diameter. Depending on
the degree of mismatch, you can lose anywhere from 1 dB to more than 10 dB. (Note
that there is no loss of this type if the sender's smaller core is connected to a larger core
at the receiving end.) This loss source is particularly bothersome for single-mode fiber,
since the cores are so small to begin with.
Connecting two fiber-optic cables whose cores are not both centered in the cladding, so
that there is spillage from the transmitter's core into the receiver's cladding.
Connecting cores, one or both of which are elliptical rather than perfectly round. Again,
this results in spillage from the sending core.
Connecting a core with a given NA (numerical aperture) to a core with a smaller NA.
Connecting two fiber-optic cables that are not properly aligned, which has the same
effect as a diameter or concentricity mismatch.
Connecting fibers that are not cut cleanly and straight at the ends. The bigger the gap, the
greater the signal loss. This potential signal loss is an excellent argument for having
the fiber cut professionally, even if you will attach the connectors.
Connecting fibers at an angle. This not only can cause signal loss, it can also cause light
to enter the second fiber at an angle different from its original path, which causes signal
distortion.
If the surface of either connector end is rough, there will not be a complete union, which
will leave space for light to escape.
If the two fibers are not actually touching, light can escape into the open area between
the fiber. This light is not only lost for the signal, but some of it can also be reflected back
into the sender's fiber. Such reflected light can interfere with signals traveling in the
proper direction.
Allowing contaminants in the connector can interfere with the connection between the
fibers.
Kinks or bends in the cable, near the connector.
Using the connector too often; that is, for too many matings, which can loosen the con-
nector and allow play between the two fibers.


Prev Page 225 Next

Connector, Fiber-Optic
213
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
Make sure all connectors in your network
are compatible. Avoid core or cladding size
mismatches if at all possible. Some mis-
matches won't work together at all; others
will introduce unnecessary signal loss.
There are quite a few different types of
fiber-optic connectors. One reason for this
is that many groups and corporations devel-
oped their own during the early days of
the technology, and most of those connec-
tor types are still around. The most common
types are described in the following sections.
An ST (straight tip) connector, developed by
AT&T, is the most widely used type of fiber-
optic connector. This type of connector is
used in premises wiring and in networks,
among other places. The connector uses a
BNC attachment mechanism, 2.5 mm fer-
rules (ceramic, steel, or plastic), and either
single-mode or multimode fiber. An ST con-
nector will last for about 1,000 matings,
except when plastic ferrules are used. In that
case, the connector is good for only about
250 matings.
Insertion loss is 0.3 dB for ceramic fer-
rules, but can be more than twice that with
plastic ferrules. A return reflection loss of
40 dB is typical with single-mode fiber.
Because this is such a widely used con-
nector type, many other connectors are com-
patible or can be made compatible with a
simple adapter. For example, adapters are
available to connect SMA to ST connectors.
Originally developed in Japan for use in tele-
communications, an FC connector uses a
threaded coupling nut for the attachment,
and 2.5 millimeter ceramic ferrules to hold
the fiber. An FC connector works with either
single-mode or multimode fiber, and will last
for about 1,000 matings.
Older style FC connectors used fibers pol-
ished to a flat surface. These connectors suf-
fered from signal distortion and loss. Newer
FC connectors use a PC polishing approach,
which applies polish to a rounded surface to
ensure physical contact between the fibers.
With PC polished fibers, FC connectors have
an insertion loss of about 0.3 dB and a
return reflection loss of around 40 dB for
single-mode fiber.
FC connectors are becoming obsolete.
They are being replaced by SC and MIC
connectors.
An SC (subscriber connector) connects two
components by plugging one connector into
the other. Once the two connectors are
latched together, they cannot be pulled apart
by sheer pressure. Instead, the connection
must be broken (for example, by pressing
a button to release a latch).
An SC connector works with either
single-mode or multimode fiber, and will
last for about 1,000 matings. It has an inser-
tion loss of 0.3 dB, and a return reflection
loss of about 40 dB.
SC connectors have replaced the older FC
and D-4 connectors used in telecommunica-
tions involving fiber-optic cable. SC connec-
tors are also becoming more popular in
networking contexts, although they are still
not nearly as popular as ST connectors for
this application.
ST Connector
FC Connector
SC Connector


Prev Page 226 Next

214
Connector, IBM Data
An MIC (medium interface connector), also
known as an FDDI connector, is a dual-fiber
connector designed by an ANSI committee
for use with fiber-optic cable in the FDDI
(Fiber Distributed Data Interface) network
architecture. The connector attaches two
fibers that help make up the two rings
specified in the FDDI architecture.
MIC connectors use a latching mecha-
nism similar to the one used for SC connec-
tors. An MIC connector works with either
single-mode or multimode fiber, and will last
for about 500 matings. It has an insertion
loss of about 0.3 dB for single-mode fiber,
and about 0.5 dB for multimode fiber.
Reflection loss is 35 dB or higher, not quite
as good as for SC connectors.
The connector is quite flexible and can be
attached either to another MIC connector,
to two ST connectors, or to a transceiver.
Because of this flexibility, MIC connectors
are becoming increasingly popular.
An SMA connector uses a threaded coupling
mechanism to make the connection. This
type of connector was originally developed
in the 1970s by the Amphenol Corporation
for use with only multimode fiber; however,
SMA connectors can now be used with
either multimode or single-mode fiber.
SMA connectors last for only about 200
matings, and they have a relatively high
insertion loss of 1.5 dB (which means about
30 percent of the signal is lost).
SMA connectors come in two forms: the
SMA-905 uses a straight ferrule, and the
MIC Connector
SMA Connector
SMA-906 uses a ferrule with a step pattern,
which is narrowest at the ferrule tip, and
widest at the back end of the ferrule.
One reason for their popularity is that
SMA connectors have been designed to meet
very stringent military specifications.
Adapters are available to connect SMA to
ST connectors.
A D-4 connector is just like an FC connec-
tor, except that the D-4 ferrule (which holds
the fiber core in place) is only 2 millimeters.
D-4 connectors can be used for single-mode
or multimode cable, and will last for about
1000 matings.
An ESCON connector is similar to the MIC
connector designed for FDDI, except that
the ESCON connector uses a retractable
cover to make it easier to attach a trans-
ceiver. The drawback is that the connection
is less robust. An ESCON connector will
last for about 500 matings, has a 0.5 dB
insertion loss, and a reflection loss of at least
35 dB.
MConnector, IBM Data
An IBM data connector is a type designed
by IBM for use in its Token Ring networks.
These connectors are used to attach a node
(or lobe) to a multistation access unit
(MAU), a wallplate, or a patch panel. MAUs
group several lobes into a ring, and may
connect to other MAUs. Patch panels serve
as wiring way stations.
D-4 Connector
ESCON Connector


Prev Page 227 Next

Connector, N-Series
215
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
The attachment mechanism is genderless, MConnector, N-Series
and involves a relatively complex mecha-
nism in which two connectors click together
to establish the connection.
An IBM data connector is self-shorting,
which means that there is a circuit across it
even if there is nothing plugged in. This is
important for maintaining the ring structure
inside a MAU.
MConnector, ISO 8877
A variant of the RJ-45 connector that is
compatible with international standards.
SEE ALSO
Connector, RJ-xx
M
Connector, MIC (Medium
Interface Connector)
A dual-fiber connector designed by an ANSI
committee for use with fiber-optic cable in
the FDDI network architecture.
SEE ALSO
Connector, Fiber-Optic
M
Connector, MMJ (Modified Modular
Jack)
A special type of modular (RJ-xx) connector,
developed by Digital Equipment Corpora-
tion (DEC) for use with its wiring scheme.
An MMJ connector uses the same snap-in
attachment mechanism as the RJ-xx connec-
tor, but the plug and the jack are keyed
(made asymmetric).
An N-series, or N-type, connector is used
with thick coaxial cable, such as in thick
Ethernet networks. N-series connectors
come in male and female versions. The con-
nection mechanism uses threads to couple
the connectors. The figure "An N-series
connector" shows an example of this type
of connector.
N-series connectors come in the following
shapes and versions:
Barrel connector: Connects two pieces of
thick coaxial cable. Each end of the
barrel connector is usually female,
which means the cable pieces must
have a male N-series connector at the
end being attached.
Elbow connector: A connector with a
right angle in it, for use in corners or
in other locations where the cabling
needs to change direction.
Terminator: Prevents a signal from
bouncing back from the end of the net-
work cable and interfering with other
signals. The terminator connects to a
AN N-SERIES CONNECTOR


Prev Page 228 Next

216
Connector, RJ-xx
male N-series connector at the end of
the trunk cable segment.
Grounded terminator: Grounds and ter-
minates a thick Ethernet trunk seg-
ment. A grounded terminator connects
to an N-series connector at the end of
a trunk cable segment, but includes a
ground cable at the end of the termi-
nator. One end of each trunk cable
segment must be grounded.
MConnector, RJ-xx
An RJ-xx connector, also known as a modu-
lar connector, comes in a plastic plug that
snaps into the appropriate socket, or jack.
RJ-xx connectors are used with twisted-pair
cable, such as for telephone cables.
The attachment mechanism involves
pushing the plug into the jack until a tooth
clicks into place to prevent the plug from
coming out.
Several RJ-xx versions are available. The
most common types are RJ-11, RJ-12, and
RJ-45. RJ-11 and RJ-12 connectors are used
with two- and three-pair (four- and six-wire)
cables. RJ-45 connectors are used with four-
pair (eight-wire) cable. Since they have eight
wires, RJ-45 connectors are larger than
RJ-11 or RJ-12 connectors.
An MMJ (modified modular jack) is a
special type of RJ-xx connector developed
by Digital Equipment Corporation (DEC)
for use with its wiring scheme. An MMJ
connector uses the same snap-in attach-
ment mechanism as the RJ-xx connector,
but the plug and the jack are keyed (made
asymmetric).
An ISO 8877 connector is a variant of the
RJ-45 connector. This type is compatible
with international standards.
M
Connector, SC (Subscriber Connector)
A type of fiber-optic connector that connects
two components by plugging one connector
into the other.
SEE
Connector, Fiber-Optic
MConnector, SMA
A fiber-optic connector type that uses a
threaded coupling mechanism to make the
connection.
SEE
Connector, Fiber-Optic
M
Connector, ST (Straight Tip)
A widely used fiber-optic connector devel-
oped by AT&T. This type of connector is
used in premises wiring and in networks,
among other places.
SEE
Connector, Fiber-Optic
M
Connector, T
A connector that generally links three pieces
of cable. Specifically, a T-connector links a
device or cable to another cable. In order to
add the linked cable, the other cable must be
spliced. The connector's name comes from
its shape.


Prev Page 229 Next

Container
217
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
M
Connector, TNC (Threaded
Nut Connector)
A connector similar to a BNC connector,
except that the TNC connector is threaded
and screws into the jack to make the connec-
tion. This type of connector is also called a
threaded Neill-Concelnan or threaded navy
connector.
M
CONS (Connection-Mode Network
Service)
In the OSI Reference Model, a network-
layer service that requires an established
connection between source and destination
before data transmission begins. The logical-
link control and media-access control sub-
layers can do error detection, flow control,
and packet acknowledgment. CONS is com-
mon in wide-area networks, and is in con-
trast to the CLNS (connectionless-mode
network service) more popular with local-
area networks.
M
Console
In a Novell NetWare environment, the mon-
itor and keyboard from which the network
administrator can control server activity is
called the console. The administrator (or
any other user, if there is a security breach)
can give commands to control printer and
disk services, send messages, and so on.
To prevent unauthorized use of the con-
sole, several steps are possible:
I Lock the console to prevent physical
access to it.
I Use the lockup feature in the Monitor
NLM (NetWare Loadable Module) to
disable keyboard entry until the user
enters the correct supervisor password.
I Use the Secure Console command to
secure the console and also to prevent
access to the debugger (which can be
used to bypass security measures).
I Be on the lookout for unauthorized
activity in the SYS:SYSTEM directory.
I Before loading an NLM, check to
make sure it is approved, which means
the module has been tested by Novell
and was found to work.
BROADER CATEGOR Y
NetWare
M
Container
A container is an element in the directory
tree for Novell's NetWare 4.x's NetWare
Directory Services (NDS). The Directory tree
contains information about all the objects
connected to all the servers in a NetWare
network or internetwork. Containers help
to group these objects into a hierarchical
structure.
A container is an object that may contain
other containers or leaf objects or both.
Within the Directory tree, a container is
allowed only below the root or below
another container, as illustrated in the figure
"An example of an NDS Directory tree."


Prev Page 230 Next

218
Contention
In an actual network, a container gener-
ally corresponds to some meaningful level of
organization or administration within the
world connected by the network, such as a
division or department of a company. A leaf
object corresponds to information about a
specific network element (node, peripheral,
user, and so on).
Two types of container objects are com-
monly used:
O (Organizations): Help to organize
and group objects in the tree. There
must be at least one organization in
a directory tree. All organizations in a
tree must be at the same level: immedi-
ately below the root.
OU (Organizational Units): Help to orga-
nize subsets of leaf objects in the tree.
OU levels are not required in a Direc-
tory tree.
BROADER CATEGOR Y
NetWare
SEE ALSO
NDS (NetWare Directory Services)
MContention
The basis for a first-come-first-serve media
access method. In a contention-based access
method, the first node to seek access when
the network is idle will get to transmit. Con-
tention is at the heart of the CSMA/CD
access method used in Ethernet networks.
Compare it with the polling and token-
passing methods.
MContext
In the CCITT's X.500 Directory Services
(DS) model, a portion of the Directory Infor-
mation Tree (DIT), which contains informa-
tion about all directory objects. In Novell's
NetWare 4.x NDS, the current location in
the Directory tree.
SEE ALSO
NDS (NetWare Directory Services)
AN EXAMPLE OF AN
NDS DIRECTOR Y TREE


Prev Page 231 Next

CORBA (Common Object Request Broker Architecture)
219
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
M
Control Character
A control character is any of several charac-
ter values that have been reserved for trans-
mission and other control functions, such as
cursor movement. For example, in the ASCII MConvergence
character set, the characters with codes
below 32 are control characters. Character 9
(Ctrl-I) is a Tab code, character 7 (Ctrl-G) is
the code for a beep, and so on.
Control characters are also known as
control codes, communication control
codes, or communication control characters. M
M
Controlled Access Unit (CAU)
SEE
CAU (Controlled Access Unit)
M
Controller
In a mainframe environment, a controller
is a device that communicates with a host
computer and mediates between this host
and the terminals accessing the host.
In a PC environment, a controller is a
device, usually a board, that is responsible
for accessing another device, and for writing
and possibly retrieving material on this
device. For example, a hard disk controller
accesses the hard disk. Controllers, also
called controller boards, mediate between
the computer and a CD-ROM or tape drive. M
The controller board generally manages the
connected device, including input and
output.
The operating system uses a controller
address to locate a disk controller. This
value is usually set directly on the controller
board, by setting jumpers or DIP switches.
M
Control Unit Terminal (CUT)
SEE
CUT (Control Unit Teminal)
A process by which network activity is
resynchronized after a change in routing;
for example, because a node was added
or dropped.
Cooperative Processing
A program execution technology that allows
different tasks in a program to be carried
out on different machines. Cooperative pro-
cessing is important for client/server com-
puting, in which an application front end
executes on a client (workstation), and a
back end executes on the server.
M
Coprocessor
A microprocessor chip that carries out a cer-
tain class of tasks on behalf of another pro-
cessor (the central processing unit, or CPU),
in order to leave the CPU available for other
work. The most commonly used coproces-
sors do floating-point arithmetic. Other
types are for graphics, disk management,
and input/output.
CORBA (Common Object Request
Broker Architecture)
A specification created by the Object Man-
agement Group (OMG) to provide a way for
applications operating in object-oriented
environments to communicate and exchange
information, even if these applications are


Prev Page 232 Next

220
Core
running on different platforms. By going
through an ORB (object request broker)
applications can make requests of objects
or other applications without knowing any-
thing about the structure of the called entity.
The ORB enables applications to commu-
nicate through an object-oriented front end, M
which makes it unnecessary to use applica-
tion- or platform-specific RPCs (remote pro-
cedure calls) to make requests or to route
and deliver responses.
In addition to ORB clients and servers,
the CORBA specification includes an IDL
(interface definition language) and APIs
(application program interfaces). The IDL
provides the ORB client with a way to spec-
ify each desired operation and any required
parameters. CORBA makes provisions for
two classes of APIs:
I A static invocation API, which can be
used to specify requests and parame-
ters in advance, so that these can be
compiled directly into the application.
I A dynamic invocation API, which must
be used to specify requests and param-
eters that will not be known until
runtime.
While CORBA version 2.0 is new,
CORBA-compliant products have been
appearing almost since the original specifica-
tion in 1992. For example, Digital's Object-
Broker software implements CORBA on a
variety of platforms including various fla-
vors of UNIX, Windows and Windows NT,
DEC OSF/1, and Macintoshes. (Object-
Broker is implemented only partially on
some of these platforms.) Microsoft is
expected to develop a competing technology
based on its OLE (Object Linking and
Embedding) standard.
PRIMAR Y SOURCE
OMG's Common Object Request Broker
Architecture specification
Core
In fiber optics, the transparent central fiber
(usually glass, but sometimes plastic)
through which a light signal travels. The
core is surrounded by cladding, which has a
lower index of refraction than the core, so
that light is reflected back into the core
when it hits the cladding.
SEE ALSO
Cable, Fiber-Optic
M
Core Gateway
On the Internet, any one of several key
routers (gateways, in older Internet termi-
nology). All networks on the Internet must
provide a path from a core gateway to the
network.
M
Corporation for Open Systems (COS)
A group concerned with the testing and
promotion of products that support the
OSI Reference Model.
MCorporation for Research and
Educational Networking (CREN)
SEE
CREN (Corporation for Research and
Educational Networking)


Prev Page 233 Next

Coupler, Fiber-Optic
221
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
M
Count to Infinity
In a distance-vector routing strategy, count
to infinity is an artifact in which certain net-
works may come to be classified as unreach-
able because routers are relying on each
others' incorrect information.
The infinity in this case refers to the dis-
tance to the network. In practice, this value
will be one more than the maximum hop
count allowed for a route. In a Novell Net-
Ware network, 16 hops (steps to the destina-
tion) would be infinite, since at most 15
hops are allowed.
BROADER CATEGOR Y
Routing, Distance-Vector
MCoupler, Fiber-Optic
Most generally, a coupler is a device for
transferring energy between two or more
channels. In fiber-optic networks, a coupler
is a device that routes an incoming signal to
two or more outgoing paths, or a device that
routes multiple incoming signals into a sin-
gle outgoing path.
Couplers are important in fiber-optic net-
works. When an electrical signal is split and
sent along parallel paths, each derived signal
is the same strength. This is not the case
with light signals. After the signal is split,
the derived optical signals are each weaker
than the original signal.
For example, if a fiber-optic coupler splits
a signal into two equal signals, each of those
derived signals is half as strong; it loses 3
decibels (dB) relative to the original signal.
Couplers can be designed to split a signal
equally or unequally.
Couplers are often described in terms of
the number of input and output signals. For
example, a 3 5 coupler has three input and
five output channels. If the coupler is bi-
directional, you can also describe it as 5 3.
Under certain conditions, particularly
when using wavelength as a basis for split-
ting or multiplexing a signal, couplers are
subject to optical crosstalk. This can hap-
pen, for example, if the wavelengths being
used are too similar, so that they are trans-
formed in similar ways by the coupler. Gen-
erally, the wavelengths used will be made
different deliberately to minimize the possi-
bility of crosstalk.
Fiber-optic couplers can be grouped in
any of several ways, based on their form and
function:
I Whether the coupler is created by
using mirrors (CSR) or by fusing fibers
(fused).
I Whether the coupler splits a signal
(splitter) or combines multiple signals
into a single one (combiner).
I Whether the coupler has its own
power supply to boost signals (active)
or simply splits signals (passive).
I Whether the coupler sends signals in
one direction (directional) or both
directions (bidirectional).
I Whether the coupler splits the signal
into two (tee) or more (star) parts.
CSR (centro-symmetrical reflective) couplers
use a concave mirror that reflects the light
from incoming fiber(s) to outgoing ones. By
CSR versus Fused Couplers


Prev Page 234 Next

222
Coupler, Fiber-Optic
adjusting the mirror, the light distribution
can be controlled.
In a fused coupler, incoming and outgoing
fibers are gathered at a central point and
wrapped around each other. By applying
heat to the wrapping point, the fibers can be
fused at this location, so that light from any
of the incoming fibers will be reflected to all
the outgoing ones.
A splitter coupler breaks a signal into multi-
ple derived signals. An important type of
splitter is a wavelength-selective coupler,
which splits an incoming signal into outgo-
ing signals based on wavelength.
In contrast, a combiner coupler, also
known simply as a combiner, combines mul-
tiple incoming signals into a single outgoing
one. A particular type of combiner is an
essential element for WDM (wavelength
division multiplexing), in which signals from
multiple channels are sent over the same
output channel. The input channels are all
transmitting at different wavelengths, and
the coupler's job is to combine the
signals in the proper manner.
An active coupler has its own electrical
power supply, which enables the coupler
to boost each of the derived signals before
transmitting it. Active couplers include elec-
trical components: a receiver that converts
the input signal into electrical form, boost-
ing capabilities, and transmitters to convert
Splitter versus Combiner Couplers
Active versus Passive Couplers
the electrical signal into an optical one
before sending it. An active coupler may also
send the signal, usually in electrical form, to
a node on a network.
A passive coupler simply splits the signal
as requested and passes the weakened sig-
nals on to all fibers. There is always signal
loss with a passive coupler.
A directional coupler can send a split signal
in only one direction. A bidirectional
coupler can send a split signal in both
directions.
A tee coupler splits an incoming signal
into two outgoing signals. This type of
coupler has three ports, and is used in
bus topologies.
A star coupler splits the signal into more
than two derived signals. Star couplers are
used in star topologies.
A passive star coupler is an optical signal
redirector created by fusing multiple fibers
together at their meeting point. This type of
coupler serves as the center of a star configu-
ration. Because the fibers are fused, a signal
transmitted from one node will be transmit-
ted to all the other nodes attached when the
signal reaches the coupler.
Passive star couplers are used for optical
(IEEE 802.4) token-bus networks that have
a passive star topology.
Directional versus Bidirectional Couplers
Tee versus Star Couplers
Passive Star Couplers


Prev Page 235 Next

Crimper
223
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
M
COW (Character-Oriented Windows)
Interface
In OS/2, an SAA (Systems Application
Architecture) compatible interface.
M
CPE (Customer Premises Equipment)
Equipment used at the customer's location,
regardless of whether this equipment is
leased or owned.
MCPI (Computer-to-PBX Interface)
In digital telecommunications, an interface
through which a computer can communicate M
with a PBX (private branch exchange).
M
CPIC (Common Programming
Interface for Communications)
APIs (Application Program Interfaces) for
program-to-program communications in
IBM's SAA (Systems Application Architec-
ture) environment. The CPIC APIs are
designed for LU 6.2 protocols; that is, for
interactions in which the programs are
equals.
M
CPU (Central Processing Unit)
The main processor in a computer. The CPU M
may be aided in its work by special-purpose
chips, such as graphics accelerators and the
UART (universal asynchronous receiver/
transmitter).
M
Cracker
Someone who tries to access computers or
networks without authorization-generally
with malicious intentions. In contrast, the
term hacker is used to refer to someone who
tries to access systems out of curiosity. The
latter term, however, is also used as a gen-
eral term for anyone trying to access a com-
puter without authorization.
MCRC (Cyclic Redundancy Check)
An error-detection method based on a trans-
formation of the bit values in a data packet
or frame.
SEE
Error Detection and Correction
CREN (Corporation for Research
and Educational Networking
Part of the Internet, along with ARPAnet,
MILnet, and several other research and
government networks.
M
CRF (Cable Retransmission Facility)
In a broadband network, the starting point
for tramsmissions to end-users. For exam-
ple, the CRF might be the cable network's
broadcast station. End-user stations can gen-
erally transmit control and error informa-
tion, but not data to the CRF.
Crimper
A tool for crimping the end of a piece of
cable in order to attach a connector to the
cable. This tool is essential if you plan to cut
and fine-tune cable.


Prev Page 236 Next

224
Cross-Connect Device
M
Cross-Connect Device
A cross-connect device is a punch-down
block. A cross-connect is a connection
between two punch-down blocks. This
device is used to establish a physical connec-
tion between the horizontal cable running
from a machine to the cable running to the
wiring center, or hub.
The device is used to terminate incoming
wire pairs in an orderly manner, and to dis-
tribute these wires to end users or to wiring
centers. By connecting a device, such as a
node in a network, to the more accessible
punch-down block instead of directly to a
wiring center or to a hub, you can switch
connections more easily; for example, to test
different wiring configurations.
MCrosstalk
Crosstalk is interference generated when
magnetic fields or current from nearby wires
interrupt electrical currents in a wire. As
electrical current travels through a wire, the
current generates a magnetic field. Magnetic
fields from wires that are close together can
interfere with the current in the wires.
Crosstalk leads to jitter, or signal distortion.
Shielding the wire and twisting wire pairs
around each other help decrease crosstalk.
If twists are spaced properly, the magnetic
fields in the wires cancel each other out.
However, crosstalk can also be induced
if the twists in a wire are badly spaced.
Crosstalk comes in near and far-end
varieties, known as NEXT and FEXT,
respectively. FEXT (far-end crosstalk) is the
interference in a wire at the receiving end
of a signal sent on a different wire. NEXT
(near-end crosstalk) is the interference in a
wire at the transmitting end of a signal sent
on a different wire. NEXT is the value gen-
erally measured when evaluating or testing
cable.
M
Cross Wye
A cable used to switch the wiring arrange-
ment from one sequence to another; for
example, from USOC wiring to EIA-568B.
This type of switch effectively changes the
pin assignments of the incoming cable.
SEE ALSO
Wiring Sequence
MCSFS (Cable Signal Fault Signature)
In electrical line testing, a unique signal
reflected back when using time domain
reflectometry (TDR) to test the electrical
activity of a line. Based on the CSFS, a
trained technician may be able to identify
the source and location of a problem.
M
CSMA/CA (Carrier Sense Multiple
Access/Collision Avoidance)
CSMA/CA is a media-access method used
in Apple's LocalTalk networks. CSMA/CA
operates at the media-access-control (MAC)
sublayer, as defined by the IEEE, of the data-
link layer in the OSI Reference Model.
When a node wants to transmit on the
network, the node listens for activity (CS,
or carrier sense). Activity is indicated by a
The CSMA/CA Process


Prev Page 237 Next

CSMA/CA (Carrier Sense Multiple Access/Collision Avoidance)
225
a
b
c
d
e
f
g
hijklmnopqrstuvwxyz
carrier on signal. If there is activity, the node
waits a period of time and then tries again to
access the network. The figure "Summary of
the CSMA/CA process" illustrates how the
method works.
The wait, known as the deferral time,
depends on the following:
I The activity level of the network. The
deferral time is longer if there is a lot
of network activity; it is shorter when
there is little activity.
I
A random value added to the base
deferral time. This ensures that two
nodes who defer at the same time do
not try to retransmit at the same time.
SUMMAR Y OF THE CSMA/CA PROCESS


Prev Page 238 Next

226
CSMA/CD (Carrier Sense Multiple Access/Collision Detect)
If the network is currently idle, the node
sends a Request To Send (RTS) signal. This
signal is sent regardless of whether the node
wants to send a directed transmission (one
with a particular destination) or a broadcast
transmission (one sent to each node on the
network).
In a directed transmission, the RTS is
addressed to a particular node, and the send-
ing node waits for a Clear To Send (CTS)
signal in reply from this node. The RTS and
the CTS must be sent within a predefined
amount of time; otherwise, the sending node M
assumes there is a collision and defers.
In Apple's LocalTalk network architec-
ture, the minimum interframe gap (IFG)-
the time between successive frames (such as
RTS and CTS or between CTS and data
transmission)-is 200 microseconds.
In a broadcast transmission, the RTS is
addressed to a predefined address (255) that
indicates broadcasts. The sending node does
not wait for a CTS; instead, the node begins
the transmission. In a broadcast transmis-
sion, the RTS serves more as a statement
of intent than as a request.
CSMA/CA is a probabilistic and contentious
access method. This is in contrast to the
deterministic token-passing and polling
methods. It is contentious in that the first
node to claim access to an idle network gets
it. CSMA/CA is probabilistic in that a node
may or may not get access when the node
tries. A disadvantage stemming from this
probabilistic access is that even critical
requests may not get onto the network in
a timely manner.
Collision avoidance requires less sophisti-
cated circuitry than collision detection, so
the chip set is less expensive to manufacture.
Collisions cannot always be avoided, how-
ever. When they occur, LocalTalk lets a
higher level protocol handle the problem.
BROADER CATEGOR Y
Media-Access Method
SEE ALSO
CSMA/CD; Polling; Token Passing
CSMA/CD (Carrier Sense Multiple
Access/Collision Detect)
CSMA/CD is a media-access method used in
Ethernet networks and in networks that
conform to the IEEE 802.3 standards.
CSMA/CD operates at the media-access-
control (MAC) sublayer, as defined by the
IEEE, of the data-link layer in the OSI Refer-
ence Model.
The following network architectures use
this access method:
I Ethernet (and 802.3 compliant
variants)
I EtherTalk, Apple's implementation
of the Ethernet standard
I G-Net, from Gateway
Communications
I IBM's PC Network, which is a
broadband network
I AT&T's StarLAN
Directed versus Broadcast Transmissions
Type of Access Method


Prev Page 239 Next

CSMA/CD (Carrier Sense Multiple Access/Collision Detect)
227
a
b
c
d
e
f
g
hijklmnopqrstuvwxyz
In CSMA/CD, a node that wants to transmit
on a network first listens for traffic (electri-
cal activity) on the network. Activity is indi-
cated by the presence of a carrier on signal
on the line. The figure "Summary of the
CSMA/CD process" illustrates how the
method works.
If the line is busy, the node waits a bit,
then checks the line again. If there is no
activity, the node starts transmitting its
The CSMA/CD Process
SUMMAR Y OF THE CSMA/CD PROCESS


Prev Page 240 Next

228
CSMA/CD (Carrier Sense Multiple Access/Collision Detect)
packet, which travels in both directions on
the network cable.
The node continues monitoring the net-
work. However, it is possible for two nodes
to both detect no activity on the line and
start transmitting at the same time. In that
case, a collision occurs, and the network
has packet fragments floating around.
When a collision is detected, a node
follows this procedure:
1. Cancels its transmission by sending a
jam signal (to indicate there is a colli-
sion and thereby prevent other nodes
from joining the fun)
2. Waits a random amount of time (the
deferral time), determined by a backoff
algorithm
3. Tries to access the network again
Internally, nodes keep track of the num-
ber of unsuccessful transmission attempts
for each packet. If this number exceeds some
predefined value, the node decides the net-
work is too busy and stops trying.
Each node in a network that uses CSMA/
CD listens to every packet transmitted. The
listener first checks whether the packet is a
fragment from a collision. If so, the node
ignores it and listens for the next packet.
If a packet is not a fragment, the node
checks the destination address. The node
will further process the packet if any of the
following is the case:
I The destination address is the node's
address.
I The packet is part of a broadcast
(which is sent to every node).
I The packet is part of a multicast and
the node is one of the recipients.
As part of this further processing, the des-
tination node checks whether the packet is
valid. (For a summary of invalid Ethernet
packets, see the section on the Ethernet
frame in the Ethernet article.)
CSMA/CD is a probabilistic, contentious
access method, in contrast to the determinis-
tic token-passing and polling methods. It is
contentious in that the first node to claim
access to an idle network gets it. CSMA/CD
is probabilistic in that a node may or may
not get access when the node tries. A disad-
vantage stemming from this probabilistic
access is that even critical requests may not
get onto the network in a timely manner.
CSMA/CD works best when most net-
work activity is light. The access method
works most poorly when the network traffic
consists of many small messages, because
nodes spend much of their time colliding,
then waiting to retransmit.
To use this access method, a node must be
able to detect network activity (carrier sense,
or CS) and to detect collisions (collision
detect, or CD). Both of these capabilities are
implemented in hardware, on board the net-
work interface card.
Because CSMA/CD is a contentious
access method, any node can access the net-
work, provided that node puts in the first
request when the network line is idle. This
makes the method multiple access (MA).
Unlike CSMA/CA, a CSMA/CD node must
be able to detect a collision on the line.
Type of Access Method


Prev Page 241 Next

CTI (Computer-Telephone Integration)
229
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
BROADER CATEGOR Y
Media-Access Method
SEE ALSO
CSMA/CA; Polling; Token Passing
M
CS-MUX (Carrier-Switched
Multiplexer)
In the FDDI (Fiber Distributed Data Inter-
face) II architecture, CS-MUX is a compo-
nent that passes time-dependent data, such
as voice or video, to the architecture's
media-access-control (MAC) layer. At that
layer, the data is handled by a special iso-
chronous media-access-control (IMAC)
component.
The CS-MUX is not part of the FDDI II
definition. Rather, the CS-MUX provides
certain types of data for FDDI. Functionally,
a CS-MUX operates at a level comparable
to the logical-link-control (LLC) sublayer of
the ISO model's data-link layer.
BROADER CATEGOR Y
FDDI (Fiber Distributed Data Interface)
MCSU (Channel Service Unit)
A CSU is part of the integrated services unit
(ISU) component that replaces a modem on
a digital line. The CSU is mainly responsible
for making the signals well-behaved and
protecting the public carrier's lines from
a malfunctioning data service unit (DSU).
In particular, a CSU prevents faulty
customer-premises equipment (CPE), such
as DSUs, from affecting a public carrier's
transmission systems and ensures that all
signals placed on the line are appropriately
timed and formed. All CSU designs must be
approved and certified by the FCC (Federal
Communications Commission).
BROADER CATEGOR Y
Digital Communications
SEE ALSO
DSU/CSU (Data Service Unit/Channel
Service Unit)
M
CTI (Computer-Telephone
Integration)
A strategy for connecting standalone or net-
worked computers to telephone switches in
such a manner that the computer can
receive, initiate, and route calls over the
switch.
There are various strategies for accom-
plishing this. For example, a special connec-
tion-a CTI link-can be used to provide a
single link between a network and a switch.
All traffic passes through the CTI link,
which may have a table or other means of
determining which client is the recipient or
initiator of a call.
Standards for CTI must be developed at
two levels: the physical and the API, or pro-
gramming, level.
I At the physical level, the rules for basic
connections between computers and
switches must be specified. For exam-
ple, a standard must specify the elec-
trical characteristics of such a connec-
tion. The CSTA (computer-supported
telecommunication applications) stan-
dard was developed by the ECMA
(European Computer Manufacturers'
Association). It has been around for a
few years, and it is being implemented


Prev Page 242 Next

230
CTS (Clear To Send)
by several vendors. A competing stan-
dard-SCAI (switch computer applica-
tions interface)-is still under
development by ANSI.
I The API level provides functions that
enable programmers to gain access to
and use the capabilities of the lower
level protocols. Little has been stan-
dardized at this level. Two widely used
APIs are Microsoft's TAPI (Telephony
Application Programming Interface)
and Novell's TSAPI (Telephony Ser-
vices API).
In addition to a CTI link, various other
elements can be introduced into a configura-
tion that integrates computers and telephony
devices and services. For example, a CTI
server can connect to the CTI link at one end
and to APIs running on network nodes at
the other end. This makes it easier to coordi-
nate and control traffic between network
and telephony services.
Data distributors, voice response units
(VRUs), and automatic call distributors can
also help make the services relying on CTI
more efficient. For example, an ACD can
help route incoming calls to the next avail-
able person in a technical support pool. As
standards for Computer Telephony become
more completely defined and accepted, we
can expect considerable activity in this area.
SEE ALSO
TAPI; TSAPI
MCTS (Clear To Send)
CTS is a hardware signal sent from a
receiver to a transmitter to indicate that
the transmitter can begin sending. CTS
is generally sent in response to a Request To
Send (RTS) signal from the transmitter. The
CTS signal is sent by changing the voltage
on a particular pin.
CTS is used most commonly in serial
communications, and is sent over pin 5 in an
RS-232 connection. The RTS/CTS combina-
tion is used in the CSMA/CA (carrier sense
multiple access/collision avoidance) media-
access method used in Apple's LocalTalk
network architecture.
BROADER CATEGOR Y
Flow Control
SEE ALSO
RTS (Request To Send)
M
CTS (Conformance Testing Service)
A series of programs developed to create test
methods for determining how well (or
whether) a product implements a particular
protocol correctly. CTS projects have devel-
oped or are developing test suites for LAN
protocols (CTS-LAN), for wide area net-
works (CTS-WAN), and for such ISO or
ITU standards as FTAM (File Transfer,
Access, and Management), X.400 (message
handling), and X.500 (directory services). In
general, the tests conform to guidelines for
abstract test suites established by the ITU.
M
CUA (Common User Access)
In IBM's SAA environment, specifications
for user interfaces that are intended to pro-
vide a consistent look across applications
and platforms.
SEE ALSO
SAA (Systems Applications Architecture)


Prev Page 243 Next

Cycle, FDDI II
231
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
M
CUT (Control Unit Terminal)
A terminal operating mode that allows only
one session, such as running an application,
per terminal. (If a CUT terminal is attached
to an IBM 3174 establishment controller
with multiple logical terminal support, it can
support multiple sessions.)
COMPARE
DFT (distributed function terminal)
M
Cut-Off Wavelength
In single-mode fiber optics, the shortest
wavelength at which a signal will take
a single path through the core.
M
Cut-Through Switching
A switching method for Ethernet networks.
The switch reads a destination address and
immediately starts forwarding packets,
without first checking the integrity of each
packet. This reduces latency.
There are two switching strategies for
implementing cut-through switches:
I Cross-bar switching, in which each
input port (segment) establishes a
direct connection with its target output
port. If the target port is currently in
use, the switch waits, which could
back packets up at the input port.
I Cell-backplane switching, in which
all ports share a common backplane
(bus) along which all packets are sent.
Incoming packets are broken up and
repackaged with target addresses.
These fragments are then sent onto the
common backplane, from which the
fragments will get themselves to the
specified output port. The backplane
should have a bandwidth at least as
high as the cumulative bandwidths
of all the ports.
COMPARE
Store-and-Forward Switching
M
CWIS (Campus-Wide Information
System)
An online repository of information about a
particular school or campus. The CWIS con-
tains information such as campus-event cal-
endars, course listings, and job openings.
Although they are created for use by stu-
dents on the individual campuses, CWISs
are accessible over the Internet.
MCycle, Periodic Analog Signal
One complete repetition of a periodic analog
signal. A cycle goes from a high point (peak)
in the signal's level to a low point (trough)
and back to the peak. The cycles per second
value defines the frequency of a periodic sig-
nal. Frequency is measured in hertz (Hz).
For example, a 50 Hz signal travels at 50
cycles per second.
M
Cycle, FDDI II
In an FDDI (Fiber Distributed Data Inter-
face) II network operating in hybrid mode,
a cycle is a 12,500-bit protocol data unit
(PDU), or packet, that provides the basic
framing for the FDDI transmission. The
cycle is repeated 8,000 times per second,
which yields 100 megabits per second
(Mbps) of bandwidth for the network.


Prev Page 244 Next

232
Cylinder
The cycle contains the following
components:
Cycle header: Specifies how the cycle is to
be used. One part of the information
specified in the 12 bytes in the header
is whether each of the wideband chan-
nels is being used for packet-switched
or isochronous data.
DPG (dedicated packet group): Used for
packet-transfer control. The DPG con-
sists of 12 bytes.
WBC (wideband channel): Used for
actual data transmission. There are 16
WBCs in each cycle. Each WBC con-
sists of 96 bytes, or octets, and may be
subdivided into subchannels. Depend-
ing on the number of bits allocated
each cycle, subchannels may have
bandwidths ranging from 8 kilobits
per second (kbps) to 6.144 Mbps. For
example, an 8-bit-per-cycle subchannel
yields a 64 kbps data rate, correspond-
ing to a B channel in the ISDN tele-
communications model; using 193
bits per cycle yields a 1.544 Mbps T1
line. The default FDDI II WBC uses all
768 bits for a single channel.
BROADER CATEGOR Y
FDDI (Fiber Distributed Data Interface)
MCylinder
On a hard disk, the term for the collection
of concentric tracks at the same position
on each of the hard disk platters.


Prev Page 245 Next

DD


Prev Page 246 Next

234
D4 Framing
D
M
D4 Framing
In digital signaling, D4 framing is a method
for identifying the individual channels in a
DS1 channel.
D4 framing groups twelve 193-bit frames M
into one D4 superframe so that each DS1
channel consists of two D4 superframes.
Within each D4 superframe, the values in
every one hundred ninety-third bit-in bits
193, 386, and so on-are used to identify
the individual (DS0) channels. Also in each
D4 superframe, the eighth bit in every chan-
nel of frames 6 and 12 is used for signal-
ing between central offices. The figure
"Elements in D4 framing" illustrates this
method.
COMPARE
ESF Framing
DA (Destination Address)
In many types of packets, a header field that
specifies the node to which the packet is
being sent. Depending on the type of address
involved, this field may be four, six, or more
bytes.
SEE ALSO
SA (Source Address)
ELEMENTS IN D4 FRAMING


Prev Page 247 Next

DAM (Data Access Manager)
235
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
M
DAA (Data Access Arrangement)
In telephony, a device required as protection M
for the public telephone network if the
user's equipment does not meet FCC
standards.
M
DAC (Digital-to-Analog Converter)
A device for converting a digital signal to an
analog one. An ADC (analog-to-digital con-
verter) changes an analog signal to a digital
signal.
M
DAC (Dual-Attachment
Concentrator)
In an FDDI (Fiber Distributed Data Inter-
face) network architecture, a concentrator
used to attach single-attachment stations
or station clusters to both FDDI rings.
M
DACS (Digital Access and
Cross-Connect System)
In digital telecommunications, a mechanism
for switching a 64 kilobit per second (kbps)
DS0 channel from one T1 line to another.
The DACS method was originally developed
for use in telephone company switching, but
it has proven useful in networking contexts.
M
Daemon
In many operating environments, a back-
ground program that begins executing auto-
matically when a predefined event occurs.
Daemons (pronounced "demons") are com-
mon in the OS/2 and UNIX environments
and are used in artificial intelligence work.
Certain terminate-and-stay resident (TSR)
programs in a DOS environment behave like
daemon programs.
Daisy Chain
A serial linkage of components, also known
as cascading. In a daisy chain, device A is
connected to device B, which is connected to
C, and so on. A daisy chain arrangement
may be used in networks based on a bus
topology. Hard drives or other devices may
be daisy chained if they are all connected to
a SCSI adapter.
M
DAL (Data Access Language)
In Macintosh-based client/server environ-
ments, an extension to the SQL database
language. DAL is intended to provide a
uniform access to any database that
supports SQL.
MDAM (Data Access Manager)
In the System 7 operating system software
for Macintoshes, DAM is a built-in capabil-
ity for accessing databases on a network.
The DAM mediates between an application
and the database being accessed.
The DAM uses database extensions to
communicate with the database. These are
database-specific system files that contain
the commands necessary to interact with
a particular database.
BROADER CATEGORIES
Macintosh


Prev Page 248 Next

236
DAMA (Demand-Assigned Multiple Access)
M
DAMA (Demand-Assigned Multiple
Access)
In telecommunications, a method for allo-
cating access to communications channels.
Idle channels are kept in a pool. When a
channel capacity is requested, an idle chan-
nel is selected, allocated the requested band-
width, and assigned to the requesting party.
M
DAN (Departmental-Area Network)
In government offices, a network that ser-
vices a single government department.
M
Dark Fiber
A term for optical fiber that has been
installed but is not being used. According to
some estimates, over 99% of the installed
fiber-optic cable is still dark fiber.
MDARPA (Defense Advanced Research
Projects Agency)
The government agency largely responsible
for the development of the ARPAnet govern-
ment/university network, which eventually
became part of the Internet. DARPA, origi-
nally known just as ARPA, is part of the U.S.
Department of Defense (DoD).
M
DAS (Disk Array Subsystem)
The carriage, cabling, and circuitry for using
multiple hard disks.
MDAS (Dual-Attachment Station)
In an FDDI (Fiber Distributed Data Inter-
face) network architecture, a station, or
node, that is connected physically to both
the primary and secondary rings. A station
can be connected directly to the ring through
a port on the DAS. In contrast, a SAS
(single-attachment station) must be
attached to a concentrator.
M
DAS (Dynamically Assigned Socket)
In an AppleTalk internetwork, a DAS is a
unique socket value, assigned, upon petition,
to a particular client.
A socket is an entity through which a pro-
gram or process, known as a socket client,
communicates with a network or with
another process. Each AppleTalk socket
is associated with an 8-bit value.
Values between 128 and 254, inclusive,
are allocated for DASs. A process running
on a node can request a DAS value. An
available value in this range is assigned
to the process. While this process is execut-
ing, the assigned value cannot be used for
another socket.
DASs are in contrast to statically assigned
sockets (SASs). SASs are allocated for use by
various low-level protocols, such as NBP
and RTMP in the AppleTalk protocol suite.
Values between 1 and 127, inclusive, are
used for SASs. Values between 1 and 63
are used exclusively by Apple, and values
between 64 and 127 can be used by what-
ever processes request the values.
BROADER CATEGOR Y
Socket


Prev Page 249 Next

Database
237
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
M
DASS (Distributed Authentication
Security Service)
DASS is a system for authenticating users
logging into a network from unattended
workstations. These workstations must be
considered suspect, or untrusted, because
their physical security cannot be guaranteed.
DASS uses public-key encryption meth-
ods, which support the more stringent
authentication methods defined in the
CCITT's X.509 specifications. In contrast
to DASS, Kerberos is a distributed authenti-
cation system that uses a private-key encryp-
tion method.
BROADER CATEGORIES
Authentication; Encryption
COMPARE
Kerberos
M
DAT (Digital Audio Tape)
A DAT is a popular medium for network
and other backups. Information is recorded
in digital form on a small audio tape cas-
sette, originally developed by Sony and
Hewlett-Packard (HP). The most common
format was a 4-millimeter tape in a helical-
scan drive, which can hold more than a
gigabyte of information.
DATs use a logical recording format
called Data/DAT. This format supports ran-
dom data reads and writes. It also allows
data to be updated in place, rather than
requiring the modified data, and perhaps
some of the unchanged data as well, to be
rewritten to a new location.
M
Data Access Language (DAL)
In Macintosh-based client/server environ-
ments, an extension to the SQL database
language. DAL is intended to provide a
uniform access to any database that
supports SQL.
MDatabase
A database is an indexed collection of infor-
mation. The index imposes an order on
the information and also provides access
to the information in the database.
The information in a database can be
accessed, modified, or retrieved using a
query language. The most widely used query
language is SQL (Structured Query Lan-
guage), which forms the basis for most other
query languages currently in use. See the
SQL article for more information about this
language.
The overwhelming majority of databases
are still text-based, rather than graphics- or
multimedia-based, but this is changing. This
development has implications, particularly
for distributed databases. Until high-speed,
long-distance telecommunications facilities
are affordable for ordinary consumers,
transmitting video over long-distance lines
will seldom be worth the price.
Database types include flat file, relational,
object-oriented, inverted-list, hierarchical,
network, and distributed.
In a flat file database, all the information
is contained in a single file. A flat file data-
base consists of individual records that are,
in turn, made up of fields. Each field may
Flat File Database


Prev Page 250 Next

238
Database
contain a particular item of information.
There is not necessarily any relationship
between records. The records are not orga-
nized in any particular way. Instead, lookup
tables are created, and these are used to find
and manipulate records.
A flat file database makes considerable
demands of a user, who may need to "pro-
gram" the required information into appro-
priate lookup tables.
NetWare versions prior to 4.x use a flat
database, called the bindery, to store infor-
mation about nodes and devices on the
network.
In a relational database, the contents are
organized as a set of tables in which rows
represent records and columns represent
fields. Certain fields may be found in multi-
ple tables, and the values of these fields are
used to guide searches. Database access and
manipulation are a matter of combining
information from various tables into new
combinations. For example, a request might
look for all records for people who work in
a particular department and whose last raise Hierarchical Database
was more than one year ago.
The overwhelming majority of databases
currently available on PCs are relational
databases. Fortunately, the theory of rela-
tional databases is well-developed, so that
robust DBMS (database management sys-
tem) packages and powerful query and
manipulation tools are available.
In an object-oriented database, the informa-
tion is organized into objects, which consist
of properties and allowable operations
involving the objects.
Objects can be defined in terms of other
objects (for example, as special cases or vari-
ants of a specific object), and can inherit
properties from such "ancestor" objects.
The Directory tree based on the information
in the NetWare Directory Services (NDS) is
an example of an object-oriented database.
In an inverted-list database, the contents
are also organized in tables, but these
tables are more content-bound (less
abstract), and therefore less easy to
manipulate and modify.
In addition to tables, an inverted-list
database also has records whose contents
help simplify certain searches. For example,
a database might have a record for each
department in a corporation, and the con-
tents of that record might be a listing of all
the employees in that department. Indexes
are used to keep track of records and to
speed access.
In a hierarchical database, the contents are
organized hierarchically, as one or more
trees. Each record in a tree has exactly one
parent and may have children. Any two
records in a hierarchical database are related
in exactly one way.
The DOS directory and file system is an
example of a hierarchical database. The
relationships involved include "is a subdirec-
tory of" and "in the same directory as."
Relational Database
Object-Oriented Database
Inverted-List Database


Prev Page 251 Next

Data Bus
239
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
A network database is similar to a hierarchi-
cal database in that there are links between
records. The main difference is that records
in a network database may have no parents
or one or more parents. This is because a
network database consists essentially of
records and links. These links do not neces-
sarily form a hierarchically organized tree.
Note that the network in this label is not
a computer network. It is a network in the
mathematical sense: elements (records) con-
nected by links (relationships).
Any of the database types can be developed
as a distributed database, because this is a
matter of database storage rather than struc-
turing. A distributed database is simply one
whose contents are stored on multiple
machines.
The fact that two employee records are
on different machines does not change the
relationship between the employees (for
example, if both work in the same depart-
ment). DBMS software will hide the distrib-
uted nature of the database from the user, so
that users need not make any adjustments to
their queries or methods for retrieving and
changing data.
M
Data Bits
In asynchronous transmissions, the bits that
actually comprise the data. Usually, 7 or 8
data bits are grouped together. Each group
of data bits in a transmission is preceded
by a start bit, then followed by an optional
parity bit, as well as one or more stop bits.
M
Data Bus
The internal bus over which devices and sys-
tem components communicate with the cen-
tral processing unit (CPU) is called a data
bus. Buses differ in their width, which is the
number of data bits that can be transported
at a time, and in their clock speed.
In general, maximum supported clock
speeds keep getting higher, with 100 mega-
hertz (MHz) speeds already available on
some processors. While processor manufac-
turers continuously leap-frog each other's
highest speeds, official bus standards change
more slowly.
In the following summaries, the quoted
clock speeds are those specified in the bus
specifications or in de facto standards. You
will be able to find faster processors than the
ones discussed.
The following bus architectures are (or have
been) popular for PCs:
ISA (Industry Standard Architecture): The
bus for the earliest PCs. Early PC ver-
sions were 8-bit and ran at 4.77 MHz;
later AT versions were 16-bit and ran
at 8 MHz.
EISA (Extended Industry Standard
Architecture): A 32-bit extension of
the ISA bus. This architecture also runs
at 8 MHz.
MicroChannel: A 32-bit proprietary
architecture from IBM, for use in most
of its PS/x and Model xx series of com-
puters. The MicroChannel bus oper-
ates at 10 MHz.
Network Database
Distributed Database
PC Data Bus Architecture


Prev Page 252 Next

240
Data Communications
VESA (Video Electronics Standards
Association): An enhanced version of
the EISA architecture, also known as
local bus. The original version was 32-
bit at 40 MHz; the newer version is
64-bit at 50 MHz.
PCI (Peripheral Component Intercon-
nect): A newer architecture from Intel,
PCI is 64-bit and operates at 33 MHz.
These bus architectures are discussed in
more detail in separate articles.
In contrast, Apple's Macintosh line of
computers has, for the most part, used the
NuBus architecture developed by Texas
Instruments. This architecture is processor-
specific, which means that it is not applica-
ble to an entire processor family.
SEE ALSO
EISA (Extended Industry Standard Archi-
tecture); ISA (Industry Standard Architec-
ture); MicroChannel; PCI (Peripheral
Component Interconnect); VESA (Video
Electronics Standards Association)
MData Communications
Data communications is the transmission of
data, commonly by electronic means, over a
physical medium. Any potentially relevant
Zen koans aside, it is generally agreed that,
to be useful, data communications require
both a sender and a receiver.
The sender and receiver are also
known as the data source and data sink,
respectively. These are connected by a data
link. The data link includes a transmission
medium (for example, wire) and the appro-
priate transmission and receiving devices
at the data source and sink. The figure
"Elements in data communications" shows
these components.
Macintosh Data Bus Architecture
Components of Data Communications
ELEMENTS IN DATA
COMMUNICATIONS


Prev Page 253 Next

Data Communications
241
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
The sender must encode and transmit
the data, and the receiver must receive and
decode the data. Data encoding may include
special treatment, such as compression to
eliminate redundancy or encryption to pre-
vent, or at least discourage, eavesdropping.
The data transmission may be any of the
following types:
Point-to-point, or direct: Over a direct
(unmediated) link between sender and
receiver. Point-to-point connections
are commonly used in small networks
and dedicated communications lines.
Mediated: Handled, and possibly modi-
fied, by intermediate stations or parties
en route to the receiver. A transmission
may be mediated simply because there
are stations between the sender and the
receiver. In such a case, all transmis-
sions take the same path.
Switched: Mediated and possibly routed
along different paths. A switched
transmission may be diverted to any
of multiple possible paths. Different
transmission elements-fixed-size
blocks, variable-sized packets, or
entire messages-can be used as the
basis for the switching.
Broadcast: Transmitted to any station or
party capable of receiving, rather than
to a specific receiver. A radio transmis-
sion is broadcast.
Types of Data Transmission
Multicast: Transmitted to any station on
a stored or specified list of addresses.
For example, electronic newsletters or
mail from special interest groups are
multicast when they are sent only to
subscribers.
Stored and forwarded: Sent to a holding
location until requested or sent on
automatically after a predefined
amount of time.
Time division multiplexed (TDM): Com-
bined with other transmissions. In this
multiplexing method, transmissions
share the entire capacity of a single
channel. For example, the transmission
might be divided into brief transmis-
sion slices that are interspersed in the
channel.
Frequency division multiplexed (FDM):
Combined with other transmissions, as
in TDM, but the multiplexed transmis-
sions split a single channel, with each
transmission taking some portion. For
example, a transmissions may use a
small frequency range within the chan-
nel's entire range.
The figure "Common data transmission
schemes" shows the most common types of
transmission.


Prev Page 254 Next

242
Data Compression
COMMON DATA
TRANSMISSION SCHEMES
The basis for the compression can be any
of the following:
I Patterns in bit sequences, as in run-
length limited (RLL) encoding
I
Patterns of occurrences of particular
byte values, as in Huffman or LZW
encoding
I Commonly occurring words or
phrases, as in the use of abbreviations
or acronyms
COMMON DATA TRANSMISSION
SCHEMES (CONTINUED)
Compression Bases
M
Data Compression
Data compression is a method of reducing
the amount of data used to represent the
original information. This can be accom-
plished by eliminating redundancy.


Prev Page 255 Next

Data Encryption Key (DEK)
243
a
b
c
d
efghijklmnopq
r
s
t
u
v
w
x
y
z
The two types of compression methods are
lossless and lossy. In lossless compression,
all the original information can be recov-
ered. Lossless methods generally compress
data to about 50 or 33 percent of the origi-
nal size. These values represent compression
ratios of 2:1 and 3:1, respectively. Lossless
compression methods rarely reach ratios
higher than 5:1 or so.
In lossy compression, some of the original
information will be lost. Lossy methods can
Compression Methods
attain compression ratios of 100:1 and even
higher.
MData Encryption Algorithm (DEA)
SEE
DEA (Data Encryption Algorithm)
MData Encryption Key (DEK)
SEE
DEK (Data Encryption Key)
COMMON DATA TRANSMISSION SCHEMES (CONTINUED)


Prev Page 256 Next

244
Data-Flow Control
M
Data-Flow Control
The fifth layer in IBM's SNA.
SEE
SNA (Systems Network Architecture)
M
Data Fork
The data fork is the data portion of a Mac-
intosh file. It is the part of a Macintosh file
that is transferred to non-Macintosh envi-
ronments, such as DOS or UNIX.
SEE ALSO
Macintosh
M
Datagram
A datagram is a packet that includes both
source and destination addresses provided
by the user, rather than by the network. A
datagram can also contain data. A message
might be sent as multiple datagrams, which
may be delivered to the destination in non-
consecutive order. Receipt of a datagram is
not acknowledged.
Datagram routing takes place at the net-
work layer of the OSI Reference Model.
Datagram transmission takes place at the
data-link layer.
Datagram services are provided in
connectionless (as opposed to connection-
oriented) transmissions. Because connection-
less transmissions do not necessarily deliver
datagrams in order, datagram services can-
not guarantee successful message delivery.
Receipt verification is the responsibility of a
higher-level protocol, which must be able to
assemble the message from the datagrams.
Protocols that provide this type of service
include UDP (User Datagram Protocol) in
the Internet's TCP/IP protocol suite, CLNP
(Connectionless Network Protocol) in the
OSI Reference Model, and DDP (Datagram
Delivery Protocol) in the AppleTalk protocol
suite.
SEE ALSO
Connectionless Service; Connection-
Oriented Service
MDatakit VCS
A data-switch product from AT&T. Datakit
VCS offers communications channels rang-
ing from 9.6 kilobits per second to 8 mega-
bits per second, and can be linked to X.25
networks.
M
Data Link
In communications, the components and
medium necessary for communication
between two stations or parties. The
medium is generally (but not necessarily)
a wire or fiber-optic cable, and the compo-
nents are the transmitting and receiving
facilities at either end of the link.
M
Data-Link Connection Identifier
(DLCI)
SEE
DLCI (Data-Link Connection Identifier)
M
Data-Link Control (DLC)
SEE
DLC (Data-Link Control)


Prev Page 257 Next

Data Protection
245
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
M
Data Network Identification Code
(DNIC)
SEE
DNIC (Data Network Identification
Code)
M
Data Over Voice (DOV)
In communications, a strategy for transmit-
ting data over the voice channel at the same
time as a voice transmission. A human lis-
tener would not hear the data being trans-
mitted. DOV requires special equipment.
SEE ALSO
DUV (Data Under Voice)
M
Data Packet
In general, a data packet is a well-defined
block that contains user or application data.
When transmitted, a data packet will also
include a considerable amount of adminis-
trative information (not data) in the packet
header and footer.
A data packet is defined for a particular
protocol. The term is also used to refer to
such packets within a particular protocol
or architecture. For example, an X.25 data
packet can contain up to 1024 bytes of
user data.
MData-PCS (Data Personal
Communications Services)
Data-PCS is a type of wireless communica-
tions service defined by Apple in a proposal
to the FCC (Federal Communications Com-
mission). The proposal was a petition to
have the FCC set aside a 40 megahertz
(MHz) bandwidth in the 140 MHz range
between 1.85 and 1.99 gigahertz (AHz).
The bandwidth is to be used for wireless
communications using radio waves. Trans-
missions within the allocated bandwidth
could have a maximum power of 1 watt.
This maximum is strong enough for a 50
meter (165 feet) transmission range, but
weak enough to allow multiple wireless
networks to operate in different parts of
the spectrum without interference.
BROADER CATEGOR Y
Transmission, Wireless
M
Data Protection
Data protection involves the safeguarding of
data being transmitted across the network
or stored somewhere on the network.
Various steps can be taken to protect net-
work data. Most of the measures cost
money, but the more steps you take, the bet-
ter protected your data is likely to be. This
article summarizes techniques for protecting
your data from equipment failures. See the
Security article for information about how
to protect data from unauthorized or mali-
cious users.
The first line of defense-at the power
lines-includes measures such as the
following:
I Make sure the outlets you are using for
the network machines are properly
grounded. Without grounding, power
protection measures may be pointless.
Protecting Data Against Power Disturbance


Prev Page 258 Next

246
Data Protection
I Use a UPS (uninterruptible power
supply) to ensure that a sudden power
sag or failure does not cause the server
or other crucial computers to crash.
When a brownout or blackout occurs,
the UPS provides emergency power
from batteries. In case of a total power
loss, the UPS should be able to power
the server long enough to permit an
orderly shutdown. A UPS can also
clean a power signal (to make it closer
to a pure waveform) before it reaches
the networking hardware.
I Use surge protectors to protect against
spikes (or surges) and sags. The former
are very short bursts of very high volt-
age; the latter are temporary drops in
voltage. When selecting surge protec-
tors, be aware that the less expensive
surge protectors are designed to pro-
tect against a single spike (or at most
against a few spikes). These protectors
are not designed to withstand repeated
spikes. More expensive protectors will
provide such long-term protection.
Make sure surge protectors and all
other electrical devices are UL listed.
I Use isolation transformers to protect
against noise and static (smaller varia-
tions in voltage). These transformers
clamp (suppress) any voltages that lie
outside a predefined range.
Other data protection measures include the
following:
I Doing regular backups, so that a mini-
mum amount of data (such as no more
than a day's worth) will ever be lost
because of system failure. See the
Backup article for more information.
I Running regular and rigorous diagnos-
tics on your hard disks. Diagnostic
programs will detect bad sectors or
sectors that are about to go bad, will
move any data from these sectors to
safe areas of the disk, and will lock out
the defective sectors. Some network
packages can do this type of redirec-
tion on the fly. See the Diagnostic Pro-
gram article for more information.
I Monitoring for viruses, and having
well-defined recovery procedures in
case of a virus attack. To reduce the
possibility of virus infections, limit
users' ability to upload software
from personal floppy disks. See the
Anti-Virus Program article for more
information.
Backups, Diagnostic,
and Anti-Virus Measures
UPS TIPS
If a UPS on every machine is too expensive, put
one on just the most crucial network compo-
nents. Make sure to protect at least the file serv-
ers. Put surge protectors on as many other nodes
as possible.
When calculating costs, keep in mind that
research has found that networks with UPSs have
lower maintenance costs than networks with just
surge protectors and isolation transformers.
Don't put a UPS on a printer. Not only is this
unnecessary, it's also futile, since the printer's
power demands will drain the UPS battery.


Prev Page 259 Next

Data Set
247
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
NetWare provides a variety of data-
protection features that can be grouped into
a category called fault tolerance. Other net-
working software may have similar features.
NetWare's fault-tolerance features include
the following:
Disk duplexing: Uses two hard disks
attached to the server, and automati-
cally copies all data to both hard disks.
The disks are each accessed through
separate channels (which means that
each disk has its own controller
board). If one disk or channel fails,
the network operating system will
notify the system administrator, and
will continue writing to the working
disk. Not all network software pack-
ages support disk duplexing.
Disk mirroring: Also uses two hard disks
and copies all data to both hard disks,
but both disks share the same channel
(which means that they are connected
to the same controller board). Failure
of the controller board makes both
disks inaccessible.
Hot Fix: Uses a special area of the hard
disk (called the redirection area) to
hold data from defective areas. When
a write operation indicates there is a
problem at the location being written,
the Hot Fix capability rewrites the
data in question to the redirection
area, and stores the address of the
defective location in a table set aside
for that purpose.
Read-after-write verification: Checks
newly written data before discarding
the source data from memory. After
writing data to the hard disk, the net-
working software reads the newly
written data and compares it with the
original data (which is still stored in
RAM). If the new data and the original
data match, the original data is dis-
carded from RAM, and the next disk
operation can take place. If there is a
discrepancy, some corrective action
(for example, a Hot Fix) is taken.
FAT duplication: Maintains duplicate file
allocation tables (FATs) and directory
entry tables (DETs). This method helps
prevent files from becoming corrupted
because of addressing errors (rather
than because of media defects). FAT
duplication is done automatically by
most networking software.
SEE ALSO
Anti-Virus Program; Backup; Diagnostic
Program; Security
M
Dataset
In some network management programs, a
term for a collection of data gathered by an
agent (a program that performs a particular
task automatically or on command). The
data will generally pertain to a particular
network function or device.
M
Data Set
In telecommunications, the telephone
company's name for a modem.
Data Protection through
Software and Hardware


Prev Page 260 Next

248
Data Set Ready (DSR)
M
Data Set Ready (DSR)
A signal from a modem, sent to indicate the
modem is ready to operate. In an RS-232C
interface, this signal is sent on pin 6.
COMPARE
DTR (Data Terminal Ready)
M
Data Sink
In data communications, the receiver of a
data transmission. This is in contrast to the
data source, which is the sender.
M
Data Source
In data communications, the sender of a
data transmission. This is in contrast to the
data sink, which is the receiver.
MData Stream Compatibility (DSC)
In IBM's SNA (Systems Network Architec-
ture), a basic, bare-bones printing mode.
COMPARE
SCS (SNA Character String)
M
Data Switch
A location or device in which data can be
routed, or switched, to its destination. Data-
switch devices are used in switching net-
works, in which data is grouped and routed
on the basis of predetermined criteria or cur-
rent network traffic.
M
Data Terminal Ready (DTR)
A signal from a modem, sent to indicate that
a device (for example, a computer) is ready
to send and receive data. In an RS-232C
interface, this signal is sent on pin 20.
COMPARE
DSR (Data Set Ready)
M
Data Transparency
Data transparency is a data-transmission
strategy designed to ensure that data will not
be interpreted as control signals. Bit or byte
sequences that might be interpreted as flags
or commands are modified before transmis-
sion and restored upon receipt.
For example, LLAP (LocalTalk Link
Access Protocol), which is used in some
AppleTalk networks, uses a data transpar-
ency method called bit stuffing to ensure
that the data bit sequence 01111110 is never
transmitted, since this specific value repre-
sents a flag. In bit stuffing, a 0 bit is inserted
after the fifth 1 value in the 01111110
sequence.
MData Warehousing
An information management strategy in
which a company's information is all acces-
sible through a single database. The corpo-
rate information may come from many
sources and departments, may come in a
variety of forms, and may be stored at dif-
ferent levels of detail. Corporate informa-
tion includes such things as product,
customer, and other "departmental" data-
bases; sales, inventory, and other transaction
data; archival, or legacy, data, and so forth.
The data warehouse will also contain
meta-data, which is information about the
general organization of the warehouse, the
format and location of the various materials


Prev Page 261 Next

dB (Decibel)
249
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
in the warehouse, the operations or uses
allowed for various items, and possibly con-
nections between data items. The meta-data
needs to be updated whenever the actual
data is changed.
The warehouse contents may be distrib-
uted over various machines and locations,
but should be accessible in a transparent
manner through a server. It is this transpar-
ent access of the entire corporate database
with simple commands that makes data
warehousing so attractive. By making the
entire database accessible, it becomes easier
to spot trends, coordinate updates, and
generally keep the data organized and
consistent.
Access to the data warehouse always
assumes user authorization. That is, the
integration of various databases should not
make it possible for users to get access to
data that were off limits before warehous-
ing. Warehouse data should be accessible to
authorized users in raw form or for analy-
ses-and the necessary retrieval and analysis
tools should be part of the data warehouse
system.
Warehouse data will vary in level of
detail, or granularity. Current data, which
is more likely to be active and in flux, will
be more detailed (finer-grained) than older
materials, which may be just summary data.
Other types of data may lie between these
two extremes.
The material in a data warehouse need
not all be online all the time. Dormant (or, at
least napping) materials may be stored on
secondary media (such as tapes or compact
discs), which may need to be mounted
before users can access them. For these
materials to belong to the data warehouse,
it's only necessary for the meta-data to
include information about these materials
and their location.
A complete data warehousing system should
have resources for:
I Defining and organizing the warehouse
contents, and storing this as meta-data
I Acquiring, displaying, and distribut-
ing data
I Managing and overseeing both the
data and the warehouse operations
I Displaying information about the
warehouse contents and organization
I Analyzing and manipulating the data
The advantages of data warehousing are
many, as are the obstacles. One of the major
issues that must be considered is how to
organize and connect very heterogeneous
information. The degree to which updates
and reorganizations can be automated will
depend strongly on the quality of the basic
organization.
M
dB (Decibel)
A decibel (abbreviated dB-from a unit
named in honor of Alexander Graham Bell)
is a tenth of a bel. It is a logarithmic unit
used to measure relative signal intensity. For
example, decibels are used to measure the
relative intensity of acoustic, electrical, or
optical signals.
A decibel value is computed by taking the
logarithm (to base 10) of a ratio, and then
multiplying this value by 10 (or 20, for some
A Data Warehousing System


Prev Page 262 Next

250
DBMS (Database Management System)
measures). For example, doubling the level
of a magnitude (such as a voltage) represents
a 3 decibel increase; conversely, halving a
level represents a 3 decibel decrease.
The decibel value may be computed in
terms of a reference level, such as a watt (W)
or a milliwatt (mW). For such measures,
the reference level is one of the values in
the ratio. These referenced measures are
denoted by dbW for decibel with reference
to one watt, and dbm for decibel with refer-
ence to one milliwatt.
MDBMS (Database Management
System)
A DBMS is application software that con-
trols the data in a database, including over-
all organization, storage, retrieval, security,
and data integrity. In addition, a DBMS
usually has the following features:
I Support for formatting reports for
printed output
I Support for importing and exporting
data from other applications using
standard file formats
I A data-manipulation language to
support database queries
SEE ALSO
Database
M
DBS (Direct Broadcast Satellite)
A satellite that broadcasts signals directly to
subscribers; that is, without going through
a central station.
M
DC (Direct Current)
Electrical power that travels in only one
direction, as opposed to alternating current
(AC), which changes directions many times
a second. Batteries and most electronic com-
ponents (such as computers) use DC power;
power supplied for homes and offices is AC.
M
DCA (Document Content
Architecture)
DCA is a data stream defined by IBM for
using text documents in various computer
environments. Three standard formats are
specified for text transfer:
RFT (Revisable Form Text): The primary
format, in which text can still be edited
FFT (Final Form Text): The format in
which text has been formatted for a
particular output device and cannot be
edited
MFT (Mixed Form Text): The format
that contains more than just text, such
as a document that also includes
graphics
COMPARE
DIA (Document Interchange
Architecture)
M
DCB (Disk Coprocessor Board)
A DCB is an expansion board that serves as
an interface between the central processing
unit (CPU) and the hard disk controller.
Because the DCB is intelligent, the CPU need
not worry about reading and writing data.


Prev Page 263 Next

DCE (Distributed Computing Environment)
251
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
A DCB is also called an HBA (host bus
adapter).
A disk channel consists of a DCB and
other components needed to connect to
one or more hard disks. Novell's NetWare
supports up to four channels. For SCSI
(Small Computer System Interface) drives,
up to eight controllers can be associated
with each DCB, and each controller can
support two hard disks.
M
DCD (Data Carrier Detect)
In telecommunications, a signal in an RS-
232 connection that is asserted (True) when
the modem detects a signal with a frequency
appropriate for the communications stan-
dard the modem is using.
M
DCE (Data Communications
Equipment)
DCE, which stands for data communica-
tions equipment or data circuit-terminating
equipment, refers to a modem that is used in
conjunction with a computer as the DTE
(data terminal equipment).
More generally, a DCE is any device
capable of communicating with the appro-
priate DTE, and of providing access to the
appropriate type of line. For example, a
modem can speak to a computer and can
provide access to analog telephone lines. In
digital telecommunications, a DSU (data ser-
vice unit) and a CSU (communications
service unit) together make up a DCE,
and provide access to the digital lines.
MDCE (Distributed Computing
Environment)
DCE is an open networking architecture
promoted by the Open Software Foundation
(OSF), which is a consortium of vendors
that includes Digital Equipment Corpora-
tion (DEC), Hewlett Packard (HP), and
IBM. The DCE architecture provides the ele-
ments needed to distribute applications and
their operation across networks in a trans-
parent fashion.
If DCE is implemented, the entire net-
work should appear to a user as one giant,
very fast and powerful computer. Regardless
of whether the network consists of two iden-
tical PCs or a few dozen different machines,
DCE protects the user from any implemen-
tation details.
DCE sits on top of whatever network
operating system is running, so that a user
interacts with the DCE environment. This
environment provides the following tools
and services for a user or an application:
I RPC (Remote Procedure Call), which
makes it possible to call an application
or function on any machine, just as if
the resource were local or even part
of the application.
I
Threads (independently executable
program segments), which can be dis-
tributed across different machines and
executed simultaneously. Threads can
speed work up considerably. The RSA
encryption algorithm-which was
expected to require over 15 years to
crack-was cracked within months
using threads.


Prev Page 264 Next

252
D Channel
I Security measures, which automati-
cally apply to the entire network.
This means that a user on a machine is
protected automatically from a virus
or unauthorized user on another
machine, just as if the intruder on the
other machine were an intruder on
that machine.
In a DCE, all nodes can be synchronized
to the DCE's clock, which effectively pro-
vides precise timing capabilities. DCE offers
both global X.500 and also local CDS (cell
directory services).
By making the entire network's resources
available in a completely transparent man-
ner, DCE helps make the fullest use of avail-
able resources, and also makes it more likely
that a resource will be available when
needed.
MD Channel
In an ISDN (Integrated Services Digital Net-
work) system, the D channel is the "data,"
or signaling, channel. The D channel is used
for control signals and for data about the
call. This is in contrast to the B channel,
which serves as a bearer for data and voice.
For BRI (Basic Rate Interface), the D
channel has a data rate of 16 kilobits per
second (kbps); for PRI (Primary Rate Inter-
face), the D channel has a data rate of 64
kbps. These two forms of the D channel are MDDBMS (Distributed Database
denoted as D16 and D64, respectively.
BROADER CATEGOR Y
ISDN (Integrated Services Digital
Network)
SEE ALSO
BRI (Basic Rate Interface), PRI (Primary
Rate Interface)
COMPARE
B Channel, H Channel
M
DCS (Defined Context Set)
In the CCITT's X.216 recommendations, an
agreed-upon context for the delivery and use
of presentation-level services.
M
DCS (Digital Cross-Connect System)
In digital telephony, a special-purpose
switch for cross-connecting digital channels
(for switching a digital channel from one
piece of equipment to another). With a DCS,
this cross-connect can take place at the rate
supported by the slower of the two lines.
MDDB (Distributed Database)
A database whose contents are stored on dif-
ferent hard disks or in different locations.
Each disk or location may be managed by
different machines. The Internet's domain
name system (DNS) is an example of a dis-
tributed database.
SEE ALSO
Database
Management System)
Database management software that can
handle a distributed database (DDB).


Prev Page 265 Next

DDP (Distributed Data Processing)
253
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
M
DDD (Direct Distance Dialing)
In telephony, the ability to dial a long-
distance number without going through
an operator.
M
DDE (Dynamic Data Exchange)
DDE is a technique for application-to-
application communications. It is available
in several operating systems, including
Microsoft Windows, Macintosh System 7,
and OS/2.
When two or more programs that sup-
port DDE are running at the same time,
they can exchange data and commands, by
means of conversations. A DDE conversa-
tion is a two-way connection between two
different applications.
DDE is used for low-level communica-
tions that do not need user intervention. For
example, a communications program might
feed stock market information into a spread-
sheet program, where that data can be dis-
played in a meaningful way and recalculated
automatically as it changes.
DDE has largely been superseded by
a more complex but more capable mecha-
nism known as Object Linking and Embed-
ding (OLE).
MDDL (Data Definition Language)
Any of several languages for describing data M
and its relationships, as in a database.
M
DDM (Distributed Data Management)
In IBM's SNA (Systems Network Architec-
ture), services that allow file sharing and
remote file access in a network.
M
DDN NIC (Defense Data Network
Network Information Center)
The DDN is a global network used by
the U.S. Department of Defense (DoD) to
connect military installations. Parts of the
DDN are accessible from the Internet, and
parts are classified.
The DDN NIC is a control center
that provides information and services
through the Internet. The DDN NIC does
the following:
I Serves as a repository for the Requests
for Comments (RFCs), which are used
to define standards, report results, and
suggest planning directions for the
Internet community.
I Assigns IP (Internet Protocol) network
addresses.
I Assigns numbers to domains (or
autonomous systems, as they are called
in the Internet jargon).
SEE ALSO
IR (Internet Registry)
DDP (Distributed Data Processing)
Data processing in which some or all of the
processing and/or I/O (input/output) work
is distributed over multiple machines.


Prev Page 266 Next

254
DDS (Dataphone Digital Service)
M
DDS (Dataphone Digital Service)
DDS is an AT&T communications service
that uses digital signal transmission over
leased lines. Because data is transmitted
digitally, no modem is required; however,
a DSU/CSU (digital service unit/channel
service unit) is needed at the interface be-
tween the digital lines and the customer's
equipment. The customer equipment will
generally be a remote bridge or router,
because DDS is commonly used for pro-
viding point-to-point links in a wide-area
network (WAN).
DDS uses four wires, supports speeds
between 2.4 and 56 kilobits per second
(kbps), and is available through most LECs
(local exchange carriers) and IXCs (inter-
exchange carriers); that is, it is available
through local or long-distance telephone
companies.
MDDS (Digital Data Service)
Leased lines that support transmission rates
between 2.4 and 56 kilobits per second.
MDE (Discard Eligibility)
In a frame-relay packet header, a bit that can
be set to indicate that the packet can be dis-
carded if network traffic warrants it. If net-
work traffic gets too heavy, the network can
discard packets that have this bit set.
M
DEA (Data Encryption Algorithm)
In general, an algorithm, or rule, for
encrypting data. In the DES, the DEA is
an algorithm for encrypting data in blocks
of 64-bits each.
SEE ALSO
DES (Data Encryption Standard)
M
DECmcc (DEC Management
Control Center)
Network management software for Digital's
DECnet networks. Products based on this
core, such as DECmcc Director, are avail-
able for specific environments.
M
DECnet
DECnet is a proprietary network architec-
ture from Digital Equipment Corporation
(DEC). DECnet has gone through several
major revisions during its lifetime. The two
most recent versions, Phases IV and V, were
released in 1982 and 1987, respectively.
Both versions are still used.
Historically, DECnet networks consisted
mainly of PDP-11s and VAXen, but the
architecture can support a broad range of
hardware, including PCs and Macintoshes.
Gateways also exist for remote access and
for access to SNA (System Network Archi-
tecture) networks.
The eight layers in the DECnet Phase IV
model correspond roughly-sometimes very
roughly-to the seven layers in the OSI Ref-
erence Model. The Phase IV layers are as
follows:
Physical: Corresponds to the OSI
physical layer. This layer establishes
a physical connection and manages
DECnet Phase IV


Prev Page 267 Next

Dedicated Circuit
255
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
the actual data transmission. This
layer supports Blue Book (as opposed
to IEEE 802.3) Ethernet protocols.
Data link: Corresponds to the OSI data-
link layer. This layer supports Blue
Book Ethernet, X.25, and DDCMP
(Digital Data Communications Mes-
saging Protocol) protocols.
Routing: Corresponds to the OSI
network layer. This layer routes
packets to their destination and helps
manage intra- and internetwork traffic.
It permits adaptive routing, gathers
network management data, and sup-
ports various routing protocols.
End-to-end communications: Corre-
sponds roughly to the OSI transport
layer. This layer helps maintain net-
work links, and segments and re-
assembles information (at sending and
receiving ends, respectively). It sup-
ports the VAX OSI Transport Service
(VOTS) protocol and DEC's own
Network Services Protocol (NSP).
Session control: Corresponds roughly
to the OSI session layer. This layer
stores network name and address
information, for use when establishing
a connection. It is also responsible for
breaking the network link when the
transmission is finished. The session
control layer supports both propri-
etary and OSI session protocols.
Network application: Corresponds
roughly to the OSI presentation layer.
This layer enables local and remote file
and terminal access. It supports OSI
presentation layer protocols and also
DEC's Data Access Protocol (DAP).
Network management: Corresponds very
roughly to part of the OSI application
layer. This layer handles peer-to-peer
network management. It supports
DEC's Network Information and
Control Exchange (NICE) protocol.
User: Corresponds very roughly to part
of the OSI application layer-the part
concerned with user applications.
DECnet Phase V was designed to comply
fully with the OSI Reference Model. This
version has only seven layers, which corre-
spond to the OSI layers. In general, DECnet
Phase V supports OSI-compliant protocols
at each level. It also supports DEC's own
protocols (such as DDCMP and DAP) for
backward-compatibility with Phase IV
networks.
Designed to handle large networks,
DECnet Phase V can use up to 20 bytes
for address information. A network can be
divided into domains for routing or adminis-
trative purposes. The address field includes
an Initial Domain Part (IDP) value, which is
unique for every network.
MDedicated Circuit
A path that goes directly from a user loca-
tion to a telephone company point of pres-
ence (POP); that is, it goes to the location at
which a subscriber's leased or long-distance
lines connect to the telephone company's
lines.
DECnet Phase V


Prev Page 268 Next

256
Dedicated Line
SEE ALSO
IXC (Interexchange Carrier), POP
(Point of Presence)
M
Dedicated Line
A dedicated line is a permanent connec-
tion-a connection that is always avail-
able-between two locations. This
connection is provided on private, or
leased, lines, rather than the public, dial-
up lines, and so a dedicated line is also
known as a leased, or private, line.
Available dedicated-line services include
the following:
DDS (Dataphone Digital Services): Pro-
vide synchronous transmission of digi-
tal signals at up to 56 kilobits per
second (kbps). Subrate (lower-speed)
services are also available, at 2,400 to
19,200 bps.
56/64 kbps lines: In Europe, these lines
provide a full 64 kbps; in the United
States and in Japan, 8 kbps are used
for administrative and control over-
head, leaving only 56 kbps for the sub-
scriber. Such lines are also available
through dial-up (nondedicated lines).
Fractional T1 lines: Lines built up in
increments of 64 kbps, to a maximum
rate of 768 kbps.
T1/E1 lines: Provide 1.544 megabits per
second (Mbps) for T1 (available in the
United States and Japan) and 2.048
Mbps for E1 (available in Mexico and
Europe) service.
The availability and pricing of these
dedicated-line services vary greatly in
different geographical areas.
COMPARE
Dial-Up Line
M
De Facto Standard
A standard that results from widespread
usage by the user community, rather than
from the work of an official standards com-
mittee. This is in contrast to a de jure stan-
dard, which gets its legitimacy from a
standards committee. De facto standards
may be just as explicitly specified as de jure
standards. De facto standards simply have
not been given a "Good Standardizing" seal
of approval. ARCnet is one of the best-
known de facto standards.
MDefault Path
In packet routing, a path used by a router to
forward a packet when the packet itself con-
tains no explicit routing instructions, and
the router has no predefined path to the
packet's ultimate destination. The default
path is generally one to a router that is likely
to have more detailed routing information.
M
Default Server
For a node, the default server is usually the
server the node logs in to. If a user is logged
in to more than one server, the default is the
server that the user is currently accessing.


Prev Page 269 Next

Delay
257
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
M
Default Value
A value used for a parameter or setting
when no other value is specified by the user
through a program or in a data file.
M
Default Zone
In an AppleTalk Phase 2 network, the zone
to which a device or node belongs until it is
assigned to a specific zone.
SEE
AppleTalk
MDeferral Time
In a CSMA (collision sense, multiple access)
media access method, the amount of time a
node waits before trying again to access the
network after an unsuccessful attempt. The
time depends on a random value and on the
network's activity level.
SEE
CSMA (Collision Sense, Multiple Access)
M
Deferred Procedure Call (DPC)
SEE
DPC (Deferred Procedure Call)
MDe Jure Standard
A standard that has been officially approved
by a recognized standards committee, such
as ANSI, CCITT, or IEEE. De jure standards
may be national or international. Popular de
jure standards include IEEE 802.3 (Ethernet)
and IEEE 802.5 (Token Ring) for networks,
and CCITT V.42bis (data compression) for
modems.
COMPARE
De Facto Standard
M
DEK (Data Encryption Key)
A value used to encrypt a message. The
DEK is used by an encryption algorithm
to encode the message, and may be used by
a decryption algorithm to decode the mes-
sage. More sophisticated encryption strate-
gies use different keys for encrypting and for
decrypting.
SEE ALSO
DES (Data Encryption Standard)
M
Delay
In an electrical circuit, a delay is a property
that slows down high-frequency signals,
causing signal distortion. An equalizer can
be used to help deal with this problem.
In a network or communications connec-
tion, a delay is a latency, or lag, before a sig-
nal is passed on or returned. This type of
delay may be due to switching or to dis-
tances involved (for example, in satellite
or cellular communications).
Some devices and connections will not
tolerate delays longer than a predefined
amount of time, and they may time-out if
this time limit is exceeded. For example, a
printer may time out if there is too long a
wait before the next instruction arrives. For
some time-sensitive devices, you can change
the default waiting time.


Prev Page 270 Next

258
Delphi Internet
M
Delphi Internet
Delphi Internet is a commercial online ser-
vice-like America Online, CompuServe, or
Prodigy. While Internet access has been the
focus of its advertisements, Delphi Internet
also offers other facilities commonly associ-
ated with online service providers. These
include forums, mail, online shopping, news
and financial information, and games and
other software to use online or to download.
Delphi Internet's Custom Forums allow
users to host and manage their own forums.
Delphi's Internet services include e-mail,
Telnet, FTP, gopher, Usenet, and IRC
(Internet Relay Chat).
FOR INFORMATION
Delphi Internet at (800) 695-4005
M
Demand-Assigned Multiple Access
(DAMA)
SEE
DAMA (Demand-Assigned Multiple
Access)
M
Demand Priority
Demand priority is a media-access method
used in 100BaseVG, a 100 megabit per sec-
ond (Mbps) Ethernet implementation pro-
posed by Hewlett-Packard (HP) and AT&T
Microelectronics. Demand priority shifts
network access control from the work-
station to a hub. This access method works
with a star topology.
In this method, a node that wishes to
transmit indicates this wish to the hub and
also requests high- or regular-priority service
for its transmission. After it obtains per-
mission, the node begins transmitting to
the hub.
The hub is responsible for passing the
transmission on to the destination node;
that is, the hub is responsible for providing
access to the network. A hub will pass high-
priority transmissions through immediately,
and will pass regular-priority transmissions
through as the opportunity arises.
By letting the hub manage access, the
architecture is able to guarantee required
bandwidths and requested service priority to
particular applications or nodes. It also can
guarantee that the network can be scaled
up (enlarged) without loss of bandwidth.
Demand priority helps increase band-
width in the following ways:
I A node does not need to keep checking
whether the network is idle before
transmitting. In current Ethernet
implementations, a wire pair is dedi-
cated to this task. By making network
checking unnecessary, demand prior-
ity frees a wire pair. This is fortunate,
because the 100BaseVG specifications
use quartet signaling, which needs four
available wire pairs.
I Heavy traffic can effectively bring
standard Ethernet networks to a stand-
still, because nodes spend most of their
time trying to access the network.
With demand priority, the hub needs
to pass a transmission on only to its
destination, so that overall network
traffic is decreased. This means there
is more bandwidth available for heavy
network traffic.


Prev Page 271 Next

DES (Data Encryption Standard)
259
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
By giving the hub control over a trans-
mission, so that the message is passed to
only its destination node or nodes, demand
priority also makes it easier to prevent
eavesdropping.
BROADER CATEGORIES
100BaseVG; Media-Access Method
M
Demarcation Point
In telephone communications, the point at
which the customer's equipment and wiring
ends and the telephone company's begins.
M
Demodulation
In communications, the process of remov-
ing and isolating the modulating signal that
was added to a carrier signal for purposes
of communication. For example, in serial
communications involving computers and
modems, the demodulation process con-
verts the acoustic signal that has traveled
over the telephone line into an electrical
form from which the transmitted data can
be determined.
M
Demultiplexer
A device that takes multiplexed material
from a single input, and sends the individual
input elements to several outputs.
M
De-osification
A term for the conversion from definitions
that conform to the OSI network manage-
ment model to definitions that conform to
the IP network management model. The
term is used in TCP/IP environments that
use SNMP (Simple Network Management
Protocol).
M
Departmental-Area Network (DAN)
SEE
DAN (Departmental-Area Network)
M
Departmental LAN
A small- to medium-sized network (up to
about 30 users) whose nodes share local
resources.
M
DES (Data Encryption Standard)
DES is the official United States data encryp-
tion standard for nonclassified documents.
DES uses a single, 64-bit value as a key and
a private-key encryption strategy to convert
ordinary text (plaintext) into encrypted form
(ciphertext). (See the Encryption article for
details on plaintext and ciphertext, as well
as private- versus public-key encryption.)
In a private-key strategy, only the sender
and the receiver are supposed to know the
key (bit sequence) used to encrypt the data.
The encryption algorithm, on the other
hand, is publicly known.
Although it is relatively difficult to crack,
DES cannot protect against fraud by the
sender or the receiver. For example, there is
no way to identify a sender who has learned
the key and is pretending to be the legitimate
sender.
An ardent early advocate for DES, the
National Security Agency (NSA) has cam-
paigned to remove DES as the official
encryption standard. The NSA is advocating
a classified algorithm (one under the NSA's


Prev Page 272 Next

260
DES (Data Encryption Standard)
control) as the basis for the encryption stan-
dard. To date, this suggestion has met with
considerable resistance from the business
and computing communities.
When the DES is used for encryption, a mes-
sage is divided into 64-bit blocks, and each
block is encrypted separately, one character
at a time. During the encryption of a block,
the computer plays an electronic shell game:
the characters in the block are scrambled 16
times during encryption, and the encryption
method changes after each scrambling. The
key determines the details of the scrambling
and the character encryption. In short, each
64-bit block goes through over a dozen
transformations during encryption.
Of the 64 bits used for the encryption
key, 56 are used for encryption, and 8 are
used for error detection. The 56 bits yield
about 70 quadrillion possible keys-almost
15 million possible keys for each person
alive today. (Imagine the key chain you
would need.)
The encryption algorithm involves several
steps:
I Permuting (switching the order of) the
bits in the block.
I Repeating a computation that uses the
data encryption key (DEK) and that
involves substitution and transposition
operations.
I Permuting the bits in the block to
restore the original order.
DEA (Data Encryption Algorithm)
DES can operate in any of four modes:
ECB (Electronic Cookbook): The sim-
plest encryption method. The encryp-
tion process is the same for each block,
and it is based on the encryption algo-
rithm and the key. Repeated character
patterns, such as names, are always
encoded in the same way.
CBC (Cipher Block Chaining): A more
involved encryption method in which
the encryption for each block depends
on the encryption for the preceding
block, as well as on the algorithm and
key. The same pattern is encoded dif-
ferently in each block.
CFB (Cipher Feedback): A still more
involved method in which ciphertext is
used to generate pseudo-random val-
ues. These values are combined with
plaintext and the results are then
encrypted. CFB may encrypt an indi-
vidual character differently each time
it is encountered.
OFB (Output Feedback): Similar to CFB,
except that actual DES output is used
to generate the pseudo-random values
that are combined with plaintext. This
mode is used to encrypt communica-
tions via satellite.
PRIMAR Y SOURCE
FIPS publication #46
BROADER CATEGOR Y
Encryption
DES Modes


Prev Page 273 Next

Device Numbering
261
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
M
Desktop
In the Macintosh environment, a file server
that provides access to applications and doc-
uments through the use of icons. On a work-
station, the desktop provides a graphical
representation of the files and programs
located on that workstation. The term also
refers to workstations that reside on users'
desks (as opposed to laptops and palmtops,
for example).
M
Destination Address
In many types of packets, the address of the
station to which the packet is being sent.
The address of the station that is sending
the message is called the source address.
M
DET (Directory Entry Table)
In Novell's NetWare, the DET is one of
two tables used to keep track of directory
information. The other table is the file allo-
cation table (FAT). The DET is stored on
a hard disk.
The DET contains information about a
volume's file and directory names and prop-
erties. For example, an entry might contain
the following:
I
File name
I File owner
I Date and time of last update
I Trustee assignments (or user rights)
I Location of the file's first block on the
network hard disk
The DET also accesses the FAT, which is
an index to the locations of the blocks that
make up each file.
The contents of the DET are stored in
special storage allocation units, called direc-
tory entry blocks (DEBs). Each DEB is 4
kilobytes, and NetWare can support up to
65,536 of these blocks.
To improve performance, NetWare can
use directory caching or hashing. Directory
caching keeps currently used directory
blocks and the FAT in a reserved area of
RAM. Frequently used directory entries will
be loaded into a cache memory. Directory
hashing is the indexing of the directory
entries, which speeds access to directory
information.
M
Device Driver
A driver program designed to enable a PC
to use or communicate with a particular
device, such as a printer or monitor. A
device driver generally has a more specific
name, such as printer driver or screen driver,
depending on the type of device involved.
M
Device Numbering
Device numbering is a method for identify-
ing a device, such as a hard disk, scanner, or
floppy drive. Three numbers serve to define
each device:
Hardware address: The address associ-
ated with the board or controller for
the device. This value is set either
through software or by setting jumpers
in the required configuration. Drivers
that need to deal with the device can


Prev Page 274 Next

262
Device Sharing
read the hardware address from the
jumper settings.
Device code: A value determined by
the location of the device's board, the
device itself, and possibly by auxiliary
components (such as controllers) asso-
ciated with the board. For example, a
device code for a hard disk includes
values for disk type, controller, board,
and disk numbers.
Logical number: A value based on the
boards to which the devices are
attached, on the controller, and on
the order in which devices are loaded.
MDevice Sharing
Use of a centrally located device by multiple
users or programs. For example, a printer
or hard disk may be shared among several
workstation users. Since most devices are
idle a high proportion of the time, sharing
them is a cost-effective way to make a
resource more widely available and more
likely to be used.
M
DFS (Distributed File System)
A file system with files located on multiple
machines, but accessible to an end-user or
a process as if the files were all in a single
location.
M
DFT (Distributed Function Terminal)
In IBM's SNA (System Network Architec-
ture), a terminal mode in which a terminal
may support up to five different sessions, so
that a user can access up to five applications
through the same terminal.
COMPARE
CUT (Control Unit Terminal)
M
DIA (Document Interchange
Architecture)
DIA is software and services defined by
IBM, to make it easier to use documents in a
variety of IBM environments. DIA includes
the following services:
I APS (Application Processing Services)
I DDS (Document Distribution Services)
I DLS (Document Library Services)
I FTS (File Transfer Service)
COMPARE
DCA (Document Content Architecture)
M
Diagnostic Program
A diagnostic program tests computer hard-
ware and peripheral devices for correct
operation. Some problems, known as hard
faults, are relatively easy to find, and the
diagnostic program will diagnose them cor-
rectly every time.
Other problems, called soft faults, can be
difficult to find, because they occur sporadi-
cally or only under specific circumstances,
rather than every time the memory location
is tested.
Most computers run a simple set of sys-
tem checks when the computer is first turned
on. The PC tests are stored in read-only
memory (ROM), and are known as power-
on self tests (POSTs). If a POST detects an
error condition, the computer will stop
and display an error message on the screen.


Prev Page 275 Next

DID (Destination ID)
263
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
Some computers will emit a beep signal to
indicate the type of error.
M
Dial-Back
In network operations, dial-back (also
known as call-back) is a security measure
to prevent unauthorized dial-up access to
a network. The networking software main-
tains a list of users and the numbers from
which they might dial in.
When a user wants to dial into the net-
work, the server takes the call, gets the user's
login information, then breaks the connec-
tion. The software then looks up the user in
the dial-up table and calls back the number
listed for the user.
As an access control and security mea-
sure, dial-back works reasonably well. How-
ever, it can fail when the user needs to dial
in from a different location, or when an
unauthorized person has gained access to
the location from which the user generally
dials in (the network calls a number, not
a person).
MDial-up Line
A dial-up line is a nondedicated communi-
cations line in which a connection can be
established by dialing the number, or code,
associated with the destination. A common
example of a dial-up line, also called a
switched line or public line, is the public
telephone line. Dial-up lines generally
support speeds of 2,400 to 9,600 bps.
The connection is created at dial-up time,
and it is destroyed when the call is finished.
This is in contrast to a leased line (also
called a private or dedicated line), in which
a connection between two specific points
is always available.
With a dial-up line, the same calling node
can be connected with an arbitrary number
of destinations. Costs accrue only for the
duration of a particular connection.
COMPARE
Dedicated Line
M
DIB (Directory Information Base)
In the CCITT X.500 Directory Services
model, the body of directory-related infor-
mation. Directory system agents (DSAs)
access the DIB on behalf of directory user
agents (DUAs).
SEE ALSO
DIT (Directory Information Tree); X.500
MDIBI (Device Independent
Backup Interface)
An interface proposed by Novell to make
it easier to move material between different
environments on the network.
M
Dibit
A pair of bits treated as a single unit. For
example, a dibit is used in certain modula-
tion methods that can encode two bits in a
single modulated value. The four possible
dibits are 00, 01, 10, and 11.
MDID (Destination ID)
In an ARCnet packet, the address of the
destination node.


Prev Page 276 Next

264
DID (Direct Inward Dialing)
M
DID (Direct Inward Dialing)
In telephone communications, a system in
which an outside caller can reach a number
in a private branch exchange (PBX) directly,
without going through a switchboard.
M
Dielectric
A nonconducting material, such as rubber or
certain types of plastic, used as an insulating
layer around the conductive wire in coaxial
and twisted-pair cable.
MDigital Access and Cross-Connect
System (DACS)
SEE
DACS (Digital Access and Cross-Connect
System)
MDigital Circuit
In communications, lines that transmit data
as unmodulated square waves, which repre-
sent 0 or 1 values. Digital circuit lines are
provided by common carriers, such as tele-
phone companies.
M
Digital Communication
Digital communication is a telecommuni-
cations method that uses digital (discrete)
signals, usually binary values, to represent
information. The original information may
be in analog or digital form.
A digital transmission uses digital, rather
than analog, signals. Digital signals are
encoded as discrete values, representing 0
or 1. These binary values may be encoded
as different voltage or current levels, or as
changes in voltage levels.
In an analog signal, information is repre-
sented as variations in a continuous wave-
form's amplitude or frequency. To transmit
analog information, the analog signal passes
through a codec (coder/decoder), which
functions as an analog-to-digital converter
(ADC). The codec samples the analog signal
thousands of times a second, representing
each sample value as a unique 8-bit digital
value.
The codec's output is a sequence of dis-
crete voltage levels, which represent the
sample values. This sequence is transmitted
over the appropriate lines, which may sup-
port speeds ranging from 2,400 bits per sec-
ond to more than 200 megabits per second.
The received digital signal is cleaned to
recover the signal information. A codec then
converts the digital signal back to analog
form. At this end, the codec serves as a
digital-to-analog converter (DAC). The sam-
pled values are used as reference points for
synthesizing a continuous waveform that
tries to reproduce the original analog signal.
The quality of the synthesized signal
depends on the sampling frequency (usually
8000 times per second) and on the number
of bits used to represent the possible signal
levels (usually 8 bits).
The elements involved in the process are
illustrated in the figure "Digital communica-
tion of an analog signal."
Compared with analog transmissions,
digital transmissions are generally less sus-
ceptible to noise, are easier to work with for
error detection and correction, and require
somewhat less complex circuitry.


Prev Page 277 Next

Digital Signal Processor (DSP)
265
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
DIGITAL COMMUNICATION
OF AN ANALOG SIGNAL
MDigital Cross-Connect System (DCS)
SEE
DCS (Digital Cross-Connect System)
MDigital ID
An element attached to an electronic mes-
sage to authenticate the message and sender.
The digital ID is assigned by a certification,
or authentication, authority, and is valid for
only a limited period. A digital ID contains
the following elements:
I The sender's name, address, and
organization
I The sender's public key
I A digital signature from the certifica-
tion authority
I A serial number for the digital ID
I Validity period for the digital ID
M
Digital Multiplexed Interface (DMI)
SEE
DMI (Digital Multiplexed Interface)
M
Digital Network Architecture (DNA)
SEE
DNA (Digital Network Architecture)
M
Digital Signal Processor (DSP)
SEE
DSP (Digital Signal Processor)


Prev Page 278 Next

266
Digital Signature
M
Digital Signature
In network security, a digital signature is a
unique value associated with a transaction.
The signature is used to verify the identity of
the sender and also the origin of the mes-
sage. Digital signatures cannot be forged.
To illustrate how digital signatures can
be used, suppose user A and user B are com-
municating using an encryption strategy,
such as the RSA public-key encryption
strategy. With the RSA strategy, user A has
a public and a private key, and user B
has a private and a public key, which
differs from user A's keys.
The figure "Communications using digi-
tal signatures and a public-key encryption
method" shows what must happen for user
A and user B to communicate using a digital
signature.
BROADER CATEGORIES
Encryption; Security Management
MDigital Speech Interpolation (DSI)
SEE
DSI (Digital Speech Interpolation)
M
Digital Termination Service (DTS)
SEE
DTS (Digital Termination Service)
M
DIP (Dual In-line Package) Switch
A DIP switch is a block with two or more
switches, each of which can be in either of
two settings. DIP switches are used as alter-
natives to jumper settings when configuring
a component. The figure "A DIP switch"
illustrates an example of a rocker-type DIP
switch.
DIP switches are used in printed circuit
boards, dot-matrix printers, modems, and
many other peripheral devices.
M
Direct Connection
In networking, a direct connection is
an unmediated connection to the network.
For example, a direct connection might
be through a network cable attached to
the network interface card (NIC).
A DIP SWITCH


Prev Page 279 Next

Direct Connection
267
a
b
c
d
efgh
ijk
l
m
n
o
p
q
r
stuv
w
x
y
z
COMMUNICATIONS USING DIGITAL SIGNATURES
AND A PUBLIC-KEY ENCR YPTION METHOD





PP
PP






PP
PP




Prev Page 280 Next

268
Direct-Control Switching
In telecommunications and wide-area net- M
works (WANs), direct connection is a con-
nection to long-distance lines that does not
go through a local carrier. This type of con-
nection is in contrast to the switched-digital
access method, in which the connection does M
go through the local carrier.
MDirect-Control Switching
In switching technology, a system in which
the path is established directly, by signals in
the network, rather than through a central
controller.
M
Directed Transmission
In an AppleTalk network using the Local-
Talk network architecture and its LocalTalk
Link Access Protocol (LLAP), a directed
transmission is one intended for a specific
node. It is in contrast to a broadcast trans-
mission, which is intended for all nodes.
In infrared communications, directed
transmission is a method in which a signal is
aimed at a central reflective target, and read
by receiving nodes as the signal bounces off
the target. This is in contrast to a diffuse
transmission, which travels in multiple
directions, but is much weaker in each
direction.
BROADER CATEGORIES
AppleTalk; Infrared Transmission; LLAP
COMPARE
Broadcast Transmission
Direct Inward Dialing (DID)
SEE
DID (Direct Inward Dialing)
Directional Coupler
A coupler that can send a split signal in
only one direction. This is in contrast to
a bidirectional coupler, which can split a
signal in more than one direction.
SEE ALSO
Coupler
M
Direct Link
A connection, or circuit, that connects two
stations directly, without any intervening
stations.
MDirectory
A directory is an organizational concept that
makes it possible to group files, so that files
can be accessed more easily. For example, all
files related to a particular project or appli-
cation may be grouped in a single directory.
To further group files, they can be placed in
subdirectories within directories.
Grouping files in a directory makes it pos-
sible to organize these files on a logical basis
and at a logical level. Creating subdirecto-
ries makes it possible to impose a hierarchi-
cal structure on files. A subdirectory is said
to be contained in a parent directory.
Grouping certain files distinguishes them
implicitly from other files that are not in the
directory. Because files in a directory are
effectively partitioned from files outside, it's


Prev Page 281 Next

Directory
269
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
possible to use the same file names in differ-
ent directories.
Directories can contain other directories,
which can contain still other directories, so
that multiple levels of containment are pos-
sible. A directory structure looks like a tree.
This tree has an infelicitously named root
directory at the top of the tree, (sub)direc-
tories as branches, and files as individual
leaves at the ends of the branches.
A file can be referred to or located by
specifying a path to it. This path consists
of a sequence of directory (or subdirectory)
names that are passed in traversing the tree
to the file. Such a path usually begins with
the root and ends with the file name.
In a file path, directory names are separated
by a special character, which differs from
environment to environment. For example,
in DOS, the separator character, or delim-
iter, is the backslash (\); in UNIX it is the
forward slash (/). Some operating environ-
ments will accept either delimiter.
In crowded or complex environments,
such as in a directory structure with many
subdirectory levels, file paths can get quite
long. Unfortunately, most operating systems
limit the number of characters allowed in a
path formulation. For example, DOS path
names can be at most 127 characters; Net-
Ware's can be up to 255 characters. Length
limitations can be a problem when trying to
pass material from one program to another.
To avoid problems with such limits, most
operating environments provide mechanisms
for specifying relative partial paths. For
example, a relative path is one that "begins"
at the current directory location (as opposed
to beginning at the root).
Versions 3.x and later of Novell's NetWare
allow you to define a subdirectory as a fake
root directory. To an application, this direc-
tory looks just like the root, and administra-
tors can assign user rights from the fake root
directory.
One advantage of a fake root is that the
real root directory need not be cluttered
because of an inflexible application. Also,
the true root directory is not compromised
because user rights must be assigned at that
level.
As stated, a directory structure is inherently
hierarchical, and can be represented as a tree
with the root at the top. This hierarchical
property can be used to keep a hard disk
organized and easy to use. It can also help
contribute to network security by making
certain types of accidents much less likely.
Directory structure refers to the way in
which directories and subdirectories are
organized in relation to each other; that is, it
refers to how they are laid out conceptually
on a hard disk or partition.
A directory structure can be flat or deep-
depending on the number of subdirectories
at the root and on the number of subdirec-
tory levels.
The Directory Hierarchy
File Path
Fake Root Directory
Directory Structure
Flat versus Deep Directory Structure


Prev Page 282 Next

270
Directory
A flat directory structure has lots of
subdirectories under the root, but few, if
any, sub-subdirectories. Such a structure is
likely to arise if there are no commonalities
in the kinds of directories being created
(and, therefore, little or no need to create
higher-level groupings). The figure "A flat
directory structure" shows an example of
this structure.
A deep directory structure, on the other
hand, may have many levels of subdirecto-
ries. For example, this type of structure
might be used if there are a few categories
of programs, with various possible activities
for these programs. The figure "A deep
directory structure" illustrates this type
of setup.
A FLAT DIRECTOR Y STR UCTURE
A DEEP DIRECTOR Y STR UCTURE


Prev Page 283 Next

Directory
271
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
In a networking context, much of the direc-
tory structure will be determined by how the
networking software sets itself up and on the
needs of users on the network. Networking
packages try to isolate system-critical files
and programs from general access. This
means that the structure will have at least
two directories: one for the system and one
for users. In practice, directory structures for
networks will be more complex than those
for stand-alone machines.
For example, Novell's NetWare creates
four predefined directories on its SYS vol-
ume: SYSTEM, PUBLIC, LOGIN, and
MAIL. Administrators and users can build
around this "proto-structure," by adding
more directories on this volume, or by creat-
ing additional volumes with different direc-
tories. UNIX-based networking software
will be installed within the existing UNIX
directory structure.
Administrators will build the files and
directories needed to run the network
around and under the predefined directories.
For example, each user may get his or her
own "home" directory, which will generally
be a subdirectory in some "user" area.
Applications should be placed in separate
directories.
When creating a directory structure and
naming directories, it is important to deter-
mine any restrictions that apply. In particu-
lar, you need to find out which application
requires the shortest file/directory names and
the shortest paths. The resulting directory
structure must be accessible even with the
most severe restrictions.
Directories are created, in part, to deal with
the proliferation of files. Similarly, partitions
on a hard disk can be created to deal with
the proliferation of directories and with the
storage requirements imposed by thousands
of files and directories.
A network file server may have to man-
age gigabytes of material-possibly more
DIRECTORY STRUCTURE
SUGGESTIONS
Various computer mavens and kibitzers have
offered suggestions about what types of directory
structures are best:
I
In terms of accessibility, for example, struc-
tures with no more than four or five levels are
recommended. With too many levels, paths
can get unacceptably long.
I
Groupings and structures should be "logical"
or "reasonable"-terms whose definitions
are generally left to the reader or the
administrator.
I
A directory should not contain "too many"
files. In some environments, the operating sys-
tem will provide at least an upper bound on
what constitutes "too many." In other cases,
the software will dictate how many files are to
be included in the directory.
I
In a network, it's often useful to structure
directories so that one set of access rights
at the top-level directory applies to all that
directory's subdirectories and files. For exam-
ple, you might put all applications in a PRO-
GRAM directory and all working files in a
WORK directory, and assign the appropriate
access rights to those two main directories.
Network Directory Structures
Higher-Level Grouping Concepts


Prev Page 284 Next

272
Directory Caching
material than can fit on a single hard disk.
To make it possible to deal with elements
at this next level of storage requirements,
higher-level grouping concepts are intro-
duced. In fact, from an information manage-
ment perspective, a file server is nothing
more than a way of grouping a few mega-
bytes of material.
Within this framework, the concept of a
directory is just a middle-level management
element. For example, in the NetWare envi-
ronment, a file is associated with:
I A file server
I A volume (which may encompass one
or more hard disks)
I Directories and subdirectories
To specify a file path, all the elements are
included, as in this example of a full Net-
Ware path:
MYSERVER/SYS:PUBLIC/INFO/TECH/
CABLE.TEX
Do not confuse the NetWare Directory
Services (NDS) Directory (which is written
with an uppercase D in Novell's documenta-
tion) with the file system directory (lower-
case d) structure maintained by the NetWare
operating system. The Directory contains
information about objects (resources, users,
and so on); the directory contains informa-
tion about files and subdirectories.
COMPARE
NDS (NetWare Directory Services)
M
Directory Caching
Directory caching is a method that uses a
fast storage area to help speed up the pro-
cess of determining a file's location on disk.
File allocation table (FAT) and directory
entry table (DET) information about the
most commonly used directory entries can
be written to the directory cache memory,
from which the information can be retrieved
quickly. Directory caching is a feature of
Novell NetWare.
The advantages of directory caching can
be augmented if the file server uses a cache
and if the requested file's contents happen
to be in the server's cache. As the directory
cache fills up, the least-used directory entries
are eliminated from the cache.
M
Directory Hashing
A method for organizing directory entries
to minimize the search time for an entry.
The hashing provides guided access to the
desired entry, so that fewer entries need
to be checked along the way.
MDirectory ID
In an AppleTalk network, a unique value
associated with a directory when the direc-
tory is created.
MDirectory Information Base (DIB)
SEE
DIB (Directory Information Base)


Prev Page 285 Next

DISA (Data Interchange Standards Association)
273
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
M
Directory Management Domain
(DMD)
SEE
DMD (Directory Management Domain)
M
Directory Rights
In various networking environments, restric-
tions and privileges that define which activi-
ties the trustee (the user or process) logged
in to the network is allowed to perform.
SEE ALSO
Access Rights
M
Directory Service (DS)
SEE
DS (Directory Service)
M
Directory Service Area (DSA)
SEE
DSA (Directory Service Area)
M
Directory Synchronization
In directory management, the task of main-
taining multiple directories, and of avoiding
or resolving inconsistencies by making sure
all directories are updated properly.
MDirectory System Agent (DSA)
SEE
DSA (Directory System Agent)
M
Directory User Agent (DUA)
SEE
DUA (Directory User Agent)
M
Direct Outward Dialing (DOD)
SEE
DOD (Direct Outward Dialing)
M
Direct Wave
In wireless communications, an electromag-
netic signal that is transmitted through the
air, but low enough to reach the destination
without being reflected off the earth or off
the ionosphere. A direct wave requires a line
of sight between sender and receiver.
M
DIS (Draft International Standard)
For international standards committees, an
early version of a proposed standard. The
DIS is circulated to all committee members
for consideration and comment.
MDISA (Data Interchange Standards
Association)
The DISA was created in 1987 to serve
as the secretariat for ASC X12 (Accredited
Standards Committee for X12), which is
the committee charged by ANSI (Ameri-
can National Standards Institute) with for-
mulating EDI (electronic data interchange)
standards. Since then, the Association has
taken on other responsibilities, including
publication of the X12 documentation, and
providing support to other standards bodies
about EDI.


Prev Page 286 Next

274
Disk Driver
M
Disk Driver
Software that serves as the interface between
the operating system and the hard disk; also
known as a disk interface driver. The net-
work vendor usually includes drivers for the
most common types of hard disks (ESDI,
SCSI, and IDE), and the hard disk manufac-
turer may include drivers for specific net-
work operating systems.
M
Disk Duplexing
Disk duplexing is a data-protection mecha-
nism that uses two or more hard disks, with
a separate channel from the PC to each disk.
(A channel is the hard disk and the compo-
nents that connect the drive to an operating
environment.) A disk-duplexing system
automatically writes everything to both
disks, using the separate channels. The
figure "Disk duplexing" illustrates this
process.
If one disk or channel fails, the network-
ing software notifies the system administra-
tor. The administrator should fix or replace
the defective disk or channel, to get it back
on line as quickly as possible. Until the disk
is replaced, the disk duplexing software will
continue writing to the working disk.
Some implementations of disk duplexing
support split seeks, in which data are read
from whichever disk finds the data first.
DISK DUPLEXING


Prev Page 287 Next

Disk Mirroring
275
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
rstuvwxyz
BROADER CATEGOR Y
Data Protection
COMPARE
Disk Mirroring
M
Disk Mirroring
Disk mirroring is a data-protection strategy
that uses two hard disks, which are accessed
through a single disk channel. (A channel is
the hard disk and the components that con-
nect the drive to an operating environment.)
All the data is written to both hard disks,
but using the same channel. The figure
"Disk mirroring" illustrates this process.
This is in contrast to disk duplexing, in
which separate channels are used.
Note that all the data is written twice in
succession with disk mirroring. Note also
that failure of the disk channel makes both
disks inaccessible.
BROADER CATEGOR Y
Data Protection
COMPARE
Disk Duplexing
IDE DRIVES AND DISK MIRRORING
IDE drives are not suitable for disk mirroring,
because one of the IDE drives is automatically
designated master and the other slave. The mas-
ter does diagnostics for both drives and controls
the slave drive. This relationship has the following
consequences, which limit the desirability of IDE
drives for disk mirroring:
I
If the master crashes, the slave is useless, since
the master runs the show for both drives.
I
If the slave crashes, the master won't find it.
Rather, the master will keep searching when
there is no response from the slave drive, and
will eventually time out.
DISK MIRRORING


Prev Page 288 Next

276
Disk Striping
M
Disk Striping
Disk striping is a data-storage strategy that
combines comparable partitions on separate
hard disks into a single volume. Data can be
read from or written to multiple partitions
at the same time, because each partition is
on a separate disk, and each disk has its own
read/write heads.
Disk striping with parity distributes par-
ity information across the partitions. If one
partition fails, the information on the other
partitions can be used to reconstruct the
missing data.
M
Disk Subsystem
The components that make up a hard disk
drive: drive unit, hard disk, controller, inter-
face card, and cable. When discussed as a
separate entity, a disk subsystem is generally
housed as an external drive.
MDISOSS (Distributed Office Supported MDistance Vector
System)
An IBM mainframe-based package that pro-
vides document preparation and electronic
mail (e-mail) capabilities.
M
Dispersion
In a fiber-optic signal, dispersion refers to
the broadening of the light signal as it trav-
els through the fiber. Dispersion is directly
proportional to distance traveled. Disper-
sion also imposes a limit on bandwidth,
because two light signals cannot become
so dispersed that they overlap.
In a wireless (infrared, radio, or micro-
wave) transmission, dispersion refers to the
scattering of the signal, which is generally
caused by the atmospheric conditions and
by any particles or objects in the transmis-
sion path.
In an electrical transmission, dispersion is
the distortion of the signal as it travels along
the wire.
M
Disruptive Test
In network management, a diagnostic or
performance test that requires a break in
ordinary network activity in order to run.
Some network management packages
require verification before running the test,
or make it possible to run such a test auto-
matically at certain times, such as when
there is little other network activity.
COMPARE
Nondisruptive Test
Distance vector refers to a class of routing
algorithms. Distance vector algorithms com-
pute distances from a node by finding paths
to all adjacent nodes and by using the infor-
mation these nodes have about continuing
on the paths adjacent to them.
Distance vector algorithms can be com-
putationally intensive, a problem that is
alleviated somewhat by defining different
routing levels.
Examples of distance vector algorithms
are the ISO's Interdomain Routing Protocol
(IDRP) and the routing information proto-
cols (RIPs) supported in the TCP/IP suite
and in Novell's IPX/SPX suite.


Prev Page 289 Next

Distributed Processing
277
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
M
Distortion
Any change in a signal, particularly, in the
signal's shape. The factors that can cause or
contribute to distortion include attenuation,
crosstalk, interference, and delay. Nonlinear MDistributed Data Processing (DDP)
distortion occurs because the signal's har-
monics (multiples of the signal's fundamen-
tal frequency) are attenuated (weakened)
by different amounts.
M
Distributed Application
A distributed application is one that exe-
cutes on multiple machines in a network,
generally, with specialized portions of the
application executing on each machine.
For example, in a client/server network,
an application front end may execute on the
user's workstation to provide an interface
for the user, and a back end for the applica-
tion may execute on a server to do the work
requested through the front end. The back
end will pass the results to the front end, and
then to the user.
This is in contrast to a centralized appli-
cation, which executes entirely on a single
machine.
M
Distributed Architecture
A configuration in which processors are
located in multiple devices, possibly in mul-
tiple locations. Each processor is capable of
functioning independently or in cooperation
with other elements in the architecture.
M
Distributed Database (DDB)
SEE
DDB (Distributed Database)
M
Distributed Data Management (DDM)
SEE
DDM (Distributed Data Management)
SEE
DDP (Distributed Data Processing)
M
Distributed File System (DFS)
SEE
DFS (Distributed File System)
M
Distributed Function Terminal (DFT)
SEE
DFT (Distributed Function Terminal)
M
Distributed Network Architecture
(DNA)
SEE
DNA (Distributed Network Architecture)
MDistributed Office Supported System
(DISOSS)
SEE
DISOSS (Distributed Office Supported
System)
M
Distributed Processing
In networking, distributed processing
describes a setup in which responsibilities
and services are spread across different
nodes or processes, so that particular tasks
are performed by specialized nodes some-
where on a network. This is in contrast to


Prev Page 290 Next

278
Distributed Relational Data Architecture (DRDA)
central processing, in which multiple nodes
share the computing power of a single
server.
In distributed processing, a station that
needs something done sends a request onto
the network. The server responsible for the
service takes on the task, does it, and returns
the results to the station. The station need
never know who actually did the work.
Distributed processing is much less sus-
ceptible to high activity levels, because the
extra work can be spread out among many
servers. On the other hand, distributed pro-
cessing requires much more extensive book-
keeping and administration, and much more
passing on of information.
COMPARE
Central Processing
MDistributed Relational Data
Architecture (DRDA)
SEE
DRDA (Distributed Relational Data
Architecture)
MDistributed System
A distributed system consists of multiple
autonomous computers that are linked
and that can-through software-give the
appearance of being a single, integrated
computer system. The individual computers
may be parts of a local, wide, or global area
network. Figure "A sample distributed sys-
tem" shows an example of such a system.
Examples of distributed systems abound,
including the Internet, various University
computing centers, and ATM (automatic
teller machine) networks.
Several features and capabilities are consid-
ered desirable for distributed systems. These
include:
I Resource sharing. This refers to
the ability for users to share hard-
ware (e.g., CPU time, peripherals),
application software (e.g., groupware),
or data (e.g., reference materials). A
resource manager can coordinate
resource allocation and sharing. Two
approaches to resource sharing are
common: client-server and object-
based. These are described more fully
below.
I Concurrency. This refers to the fact
that multiple users may be requesting
or accessing system resources at the
same time. Ideally, processors should
be able to deal with multiple users
simultaneously. A distributed system
automatically demonstrates concur-
rency each time two or more users do
things at the same time on their own
machines.
I Openness. An open system is one for
which specifications and interfaces
have been made public, so that devel-
opers can create products for the sys-
tem. An open system can more easily
handle new hardware or software con-
figurations because there are officially
accepted specifications. Open systems
also adhere to open principles for
internal operations. For example,
IPC (interprocess communication) calls
provide a standard mechanism for pro-
cesses or components to communicate
with each other.
Features of Distributed Systems


Prev Page 291 Next

Distributed System
279
a
b
c
d
efghijklmnop
q
r
s
t
u
v
w
x
y
z
I Transparency. This refers to the fact
that a user doesn't need to know that
different resources being used may
be scattered all around the world.
For the user, there should be no signi-
ficant difference between requesting a
local resource and one at some remote
location.
I Scalability. This refers to the ability
of the system to grow-for example,
through the addition of new comput-
ers or by the creation of internetworks.
When a distributed system grows,
certain information may need to be
duplicated at multiple locations in
order to maintain the efficiency of the
original, smaller system. Such replicas
must be updated and corrected in a
synchronized manner.
I Fault tolerance. This refers to the sys-
tem's ability to continue functioning
after one or more components become
unavailable because of either hardware
or software failure. One way to handle
hardware failure is to include redun-
dant components in the system. This is
A SAMPLE DISTRIBUTED SYSTEM


Prev Page 292 Next

280
Distribution Frame
an effective but expensive solution. In
a fault tolerant system, a software fail-
ure will affect only the process or pro-
cesses that failed. Among other things,
this means that a process should not be MDistribution Frame
able to freeze another process or over-
write the memory or data for another
process.
In a client-server approach to resource
sharing, each server process is a centralized
resource manager-that is, transactions
generally go through a server. Servers may
provide only certain services, and may com-
plement each other with respect to the ser-
vices they provide. A client-server approach
works well for general-purpose sharing of
information and resources.
In an object-based approach, each
resource is regarded as an object that can
be moved anywhere in a distributed system
while still remaining accessible. In an object-
based approach, all shared resources can be M
viewed in the same way. An object manager
can control access to objects or classes of
objects.
An important task in a distributed sys-
tem is the handling of the file and directory
system. Various approaches have been devel-
oped for this task. These include the Net-
work file system (NFS) from Sun and the
Andrew File System (AFS) from Carnegie-
Mellon University. Currently, distributed
systems are most likely to use UNIX
machines, partly because useful file
systems have been developed for UNIX
environments.
Distributed systems are in contrast to cen-
tralized systems in which multiple users may
be connected via terminals or PCs to a single
host machine, which may itself be a PC.
Mainframe-based centralized systems are
sometimes known as monolithic systems.
A location at which wiring is concentrated.
In a sub- or intermediate distribution frame
(SDF or IDF), wiring from components
(such as nodes in a network) is concentrated
at a single location. A backbone cable runs
from such SDFs to the main distribution
frame (MDF), which serves as a wiring cen-
ter for all the voice and data cable in a build-
ing, and which connects the building to the
larger power structures in the outside world.
M
Distribution List (DL)
SEE
DL (Distribution List)
DIT (Directory Information Tree)
In the CCITT's X.500 Directory Services
(DS) model, a directory information tree
(DIT) contains the information for a direc-
tory information base (DIB).
The information in a DIT will generally
be distributed. This provides faster access to
the information at the distributed locations.
Since a DIT can get quite large, distributing
it also helps keep down the size of the DIT
materials at any single location.
The objects in a DIT may represent inter-
mediate categories, such as country, organi-
zation, or organizational unit, or they may
represent specific objects, such as a device,
Objects in a DIT


Prev Page 293 Next

Diversity
281
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
a person, or an alias for either of these. The
root of the DIT is an imaginary entry with a
null name. This serves as a base for naming
elements in the tree.
An object gets its name from the path
between the tree's root and the object. A
particular object may be found in multiple
locations in the tree; that is, the object may
have multiple names. For example, a partic-
ular end-user might be found in the DIT as
a CPA by day (on a path through the user's
employer) or as a rock guitarist by night
(on a path through a musician's union).
A DIT does not contain the actual
objects, just information about them.
Each location in the tree has predefined
attributes associated with it. The attributes
will depend on the object class to which
the entry belongs. An object class, such
as country or organization, determines
which attributes are mandatory and which
are optional for objects belonging to
that class.
Objects in the tree will have specific
values associated with these attributes.
Although an object may appear at multiple
locations in the DIT, each object will have
only one body of information associated
with it.
Two general classes of operations are possi-
ble in a DIT: retrieval (reading) and modifi-
cation (creating and writing). A given DIT
operation may apply to a single entry or to a
group of entries. The X.500 model supports
three of the four possible operation classes:
I Retrieve a single entry
I Retrieve a group of entries
I
Modify a single entry
The fourth operation class, Modify a
group of entries, is not supported in X.500.
End-users or processes can access the infor-
mation in the DIT as follows:
I A directory user agent (DUA) pro-
vides the user with access to the DIT
through an access point. A particular
access point may support one or more
of the operation classes.
I A directory system agent (DSA) pro-
vides the requested services for the
DUA, and can also provide services
for other DSAs. Since the DIT can be
large and may be distributed, more
than one DSA may be involved. A par-
ticular DSA is generally responsible for
a portion of the DIT. This portion is
known as a context.
BROADER CATEGOR Y
X.500
MDiversity
In microwave communications, diversity
refers to either of two strategies for provid-
ing safeguards against equipment failure:
Frequency diversity: A separate fre-
quency band is allocated for use in
case the main band cannot be used
(for example, because of noise or
other interference).
Operations on the DIT
Using the DIT


Prev Page 294 Next

282
DIX (Digital Intel Xerox)
Space diversity: Two receiving anten-
nas are set up close-but not too
close-to each other. If the primary
target antenna malfunctions, the
auxiliary antenna will be used to
pull in the signals.
MDIX (Digital Intel Xerox)
The three companies whose early work on
networking eventually led to the develop-
ment of the Blue Book Ethernet standard.
M
DL (Distribution List)
In the 1988 version of CCITT's X.400
Message Handling System (MHS), a tool
for reaching multiple recipients with a single
transmission. The DL includes all addresses
to which a message is to be sent.
M
DLC (Data Link Control)
As a general term, DLC refers to the func-
tions provided at the data-link layer of the
OSI Reference Model. These functions are
generally provided by a logical-link-control
(LLC) sublayer.
SEE ALSO
Protocol, DLC
MDLCI (Data Link Connection
Identifier)
In frame-relay communications, a field
in the frame-relay header. The DLCI rep-
resents the virtual circuit number asso-
ciated with a particular destination.
M
DLL (Dynamic Link Library)
A DLL is a precompiled collection of exe-
cutable functions that can be called in pro-
grams. Instead of linking the code for called
DLL functions into a program, the program
merely gets a pointer to the DLL at runtime.
The required DLL file must be accessible at
runtime, however. Multiple programs can
use the same DLL.
DLLs are used extensively in Microsoft
Windows, OS/2, and in Windows NT. DLLs
may have file-name extensions of .DLL,
.DRV, or .FON.
M
DLS (Data-Link Services)
The services provided at the data-link layer
in the OSI Reference Model.
M
DMA (Direct Memory Access)
Direct memory access is a method for trans-
ferring data from a drive or other peripheral
device directly to the computer's memory,
without involving the CPU (central process-
ing unit).
The DMA process is managed by a spe-
cialized DMA controller chip, which is gen-
erally faster than the processor. When the
data transfer is finished, the controller chip
informs the processor, which can then pro-
ceed as if the processor had managed the
transfer. Each DMA controller can handle
up to four devices.


Prev Page 295 Next

DNIS (Dialed Number Identification Service)
283
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
M
DMD (Directory Management
Domain)
In the CCITT's X.500 Directory Manage-
ment Services, a collection of one or more
directory system agents (DSAs), and possibly
of some directory user agents (DUAs), all
managed by a single organization.
SEE ALSO
X.500
M
DMI (Desktop Management Interface)
DMI provides a standard method for identi-
fying PC hardware and software compo-
nents automatically, without intervention
from the user. At a minimum, DMI identifies M
the following information about any compo-
nent installed in a PC:
I Manufacturer
I Component name
I Version
I Serial number (if appropriate)
I Installation time and date
DMI is supported by Digital Equipment
Corporation (DEC), IBM, Intel, Microsoft,
Novell, Sun, and more than 300 other
vendors.
M
DMI (Digital Multiplexed Interface)
In digital telecommunications, a T1 interface
between a private branch exchange (PBX)
and a computer.
M
DNA (Digital Network Architecture)
A layered architecture from Digital Equip-
ment Corporation (DEC). DNA is imple-
mented in the various incarnations of
DECnet.
SEE ALSO
DECnet
MDNA (Distributed Network
Architecture)
A term for a network in which processing
capabilities and services are distributed
across the network, as opposed to being
centralized in a single host or server.
DNIC (Data Network Identification
Code)
A unique, four-digit value assigned to public
networks and to services on those networks.
MDNIS (Dialed Number Identification
Service)
A telephony service that retrieves informa-
tion about the number being called. This
information can include the name of the
number's owner and the number's location.
DNIS is very commonly used with 800 and
900 lines. For example, when multiple
lines-each with different numbers-all
come into the same call distributor, DNIS
can tell which number a caller used.
COMPARE
ANI (Automatic Number Identification)


Prev Page 296 Next

284
DNS (Domain Naming System)
M
DNS (Domain Naming System)
DNS is the distributed naming service used
on the Internet. The DNS can provide a
machine's IP address, given domain names
for the machine. Various products have
been developed to provide DNS, such as the
Berkeley Internet Name Domain (BIND).
DNS is described in RFCs 1101, 1183,
and 1637.
The basis for the domains in the DNS
may be geographical, such as an entire coun-
try, or organizational, such as a common
group or activity. The top-level domains rep-
resent the most general groupings, and these M
domain names are standardized. There are
currently 7 top-level organizational domains
and 59 top-level geographical domains. See
the tables "Internet Top-Level Organization
Domains" and "Internet Top-Level Geo-
graphic Domains" for lists of these domains.
An Internet name consists of a userid fol-
lowed by an at sign (@), which is followed
by one or more names separated by dots.
The most general of these names refers to
domains. Domain names are found at the
end of an Internet name.
A particular name may include references
to one or more domains. The rightmost of
these is a top-level domain. The ordering
from specific to general in an Internet name
is in contrast to the elements in an IP
(Internet Protocol) address, in which
the first (leftmost) number represents
the most general division.
DOAM (Distributed Office
Applications Model)
DOAM is an overarching OSI (Open Sys-
tems Interconnection) model for several
application-layer processes. The DOAM
deals with document and data organization
and transmission. Its functions include the
following:
I Document Filing and Retrieval (DFR)
I Document Printing Application (DPA)
I Message-Oriented Text Interchange
System (MOTIS)
I Referenced Data Transfer (RDT)
M
Document Management
Document management refers to the range
of tasks and considerations that may arise in
relation to the online creation, modification,
and storage of simple, compound, or hyper-
text documents.
Internet Domains
INTERNET TOP-LEVEL
ORGANIZATIONAL DOMAINS
DOMAIN NAME
INTERPRETATION
com
edu
gov
int
mil
net
org
Commercial organization
Educational institution
Government agency or
organization
International organization
U.S. military
Networking organization
Nonprofit organization
Domain Names in Internet Addresses


Prev Page 297 Next

Document Management
285
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
I A simple document contains text and
possibly formatting commands, but no
graphics, voice, etc.
I A compound document-also known
as a multimedia document-can
include graphics, sound or video, in
addition to text.
I A hypertext document is one that con-
tains links to other documents or other
locations in the same document. With
INTERNET TOP-LEVEL GEOGRAPHICAL DOMAINS
DOMAIN
NAME
INTERPRE-
TATION
DOMAIN
NAME
INTERPRE-
TATION
DOMAIN
NAME
INTERPRE-
TATION
aq
ar
at
au
be
bg
br
ca
ch
cl
cn
cr
cs
de
dk
ec
ee
eg
es
fi
Antarctica
Argentina
Austria
Australia
Belgium
Bulgaria
Brazil
Canada
Switzerland
Chile
China
Costa Rica
Czech and
Slovak Republics
Germany
Denmark
Ecuador
Estonia
Egypt
Spain
Finland
FR
GB
GR
HK
HR
HU
IE
IL
IN
IS
IT
JP
KR
KW
LI
LT
LU
LV
MX
MY
France
Great Britain
Greece
Hong Kong
Croatia
Hungary
Ireland
Israel
India
Iceland
Italy
Japan
South Korea
Kuwait
Liechtenstein
Lithuania
Luxembourg
Latvia
Mexico
Malaysia
nl
no
nz
pl
pr
pt
re
se
sg
si
su
th
tn
tw
uk
us
ve
yu
za
Netherlands
Norway
New Zealand
Poland
Puerto Rico
Portugal
Reunion
Sweden
Singapore
Slovenia
Soviet Union
Thailand
Tunisia
Taiwan
United Kingdom
United States
Venezuela
Yugoslavia
South Africa


Prev Page 298 Next

286
Document Management
the appropriate software, a user can
access the material associated with
such links from within the document.
Hypertext documents may be simple
or compound. The materials accessible
through a hypertext document may be
located in different places. For exam-
ple, the material accessible from a
home page on the World Wide Web
(WWW) might be located on machines
scattered all around the world.
Tasks such as the following are consid-
ered part of document management. Note
that in some cases the required tools are
generic, and are not tied to document
management systems. For example, encryp-
tion or compression programs are used for
purposes other than document management.
I Creation. Documents may be created
in many different ways: by scanning
existing documents for text (and possi-
bly also for graphics), with an ordinary
text editor, word processor, desktop
publishing program, or hypertext
(e.g., HTML) editor. Depending on the
method used to create the document,
the result may be a simple or a com-
pound one.
I Storage. A document can be stored as
one or more elements. The media on
which a document is to be stored may
be considered primary, secondary, or
tertiary. Primary media are those that
are almost always available and very
frequently used. Hard disks are the
best example of a primary medium.
Secondary media are also almost
always available, but have much
slower access times than primary
media. CD-ROM drives are a good
example of secondary media. Tertiary
media are available only upon request,
and they usually have slower access
times than primary media. Tapes or
discs that must first be mounted are
examples of tertiary media.
I
Retrieval. Users must be able to call
up and view documents. Ideally, the
online view of a retrieved document
should be comparable to a printed ver-
sion. That is, formatting and layout
information should be preserved. This
requires the use of special viewers or
browsers that can interpret the format-
ting and layout commands and can
translate them into the appropriate
display instructions. Popular viewers
include Acrobat from Adobe, World-
View from Interleaf, and DynaText
from Electronic Book Technologies.
I Transmission. To be truly useful, a
document management system must be
accessible to multiple users. These may
be in different geographical locations.
Consequently, it may be necessary to
send a document from one location to
another. The transmission should be as
efficient and inexpensive as possible,
but should be error-free, and should
leave the document unchanged.
I Reception. Just as it must be possible
to send a document to specified loca-
tions, it must also be possible to
receive the document at that location.
Resources must be available to recon-
struct the document (for example, if it


Prev Page 299 Next

Domain
287
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
was sent in packets) and to check its
integrity.
I Revision. Very few documents are per-
fect right from the start. As a result,
users must be able to revise docu-
ments. For simple documents, this can
be done using a text editor; for com-
pound documents, more sophisticated
editing capabilities are needed. Editors
that can use markup languages such
as HTML (HyperText Markup Lan-
guage) or its more general and pow-
erful predecessor SGML (Standard
Generalized Markup Language) are
becoming increasingly popular.
I Compression. Compression reduces
a document's size by taking advantage
of redundancy in the document. This
saves storage and also saves money
when the document is transmitted.
Compression of compound docu-
ments can get complicated since differ-
ent types of compression algorithms
are most appropriate for text and
images.
I Encryption. Encryption makes a docu-
ment more difficult to use if stolen-
since the document will be gibberish
to anyone who doesn't know the
encryption method or key. Docu-
ment encryption is particularly impor-
tant with personal and financial data.
Encryption and compression are often
used together. In such cases, it's
extremely important to do things in
the correct order. For example, com-
pressing and then encrypting is most
effective for text documents. If such
a document is transmitted, the algo-
rithms must be applied in reverse order
at the receiving end-that is, decryp-
tion then decompression.
Document management software can be
grouped into three categories:
I File managers, which generally work
with only a single or a limited num-
ber of file formats. During storage,
documents may be converted to the
supported format, which may be
proprietary.
I Library managers, which handle
documents in their native formats
and which include security capabilities.
Library managers can also track docu-
ment versions.
I Compound Document Managers,
which treat documents as virtual enti-
ties that are always subject to change.
Instead of handling a document as a
static object, a compound document
manager sees a document more as a set
of pointers to various elements, any of
which may be revised between one
viewing and the next.
M
DOD (Direct Outward Dialing)
In a Centrex or a private branch exchange
(PBX), a service that makes it possible to get
an outside line directly, without going
through the system's switchboard.
M
Domain
In both the Internet and OSI (Open System
Interconnection) communities, the term


Prev Page 300 Next

288
Domain Specific Part (DSP)
domain refers to an administrative unit. The
details of such a unit, however, differ in the
Internet and OSI environments.
In the Internet community, a domain is an
element in the DNS (Domain Naming Sys-
tem), which is a naming hierarchy. See the
DNS article for more information about
Internet domains.
In the OSI community, a domain is also a
division created for administrative purposes. MDOS Client
In this context, the details are based on func-
tional differences. The five management
domains defined in the OSI model are
accounting, configuration, fault, perfor-
mance, and security. See the Network Man-
agement article for more information about
these domains.
The term has several other meanings in
different networking contexts:
I In IBM's SNA (Systems Network
Architecture), a domain represents all
the terminals and other resources con-
trolled by a single processor or proces-
sor group.
I In Novell's NNS (NetWare Name
Service), the collection of servers that
share bindery information constitutes
a domain.
I In NetWare 4.x, a domain is a special
area in which an NLM (NetWare
Loadable Module) can run.
NetWare 4.x actually has two domains
for NLMs: OS_PROTECTED and OS. In
the OS_PROTECTED domain, you can run
untested NLMs to ensure that they do not
corrupt the operating system memory. The
OS domain is where NLMs that are proven
reliable can run more efficiently.
SEE ALSO
DNS (Domain Naming System); Network
Management
MDomain Specific Part (DSP)
SEE
DSP (Domain Specific Part)
A workstation that boots DOS and gains
access to the network using workstation
software.
M
DOS Extender
Software that enables DOS programs to
execute in protected mode, and to make
use of extended memory. Two widely
used DOS extender specifications are VCPI
(Virtual Control Program Interface) and
DPMI (DOS Protected Mode Interface).
SEE ALSO
DPMI, Protected Mode, VCPI
MDOS Requester
In Novell's NetWare 3.12 and 4.x, the
DOS Requester is client software that runs
on a workstation and mediates between
applications, DOS, and NetWare. The DOS
Requester replaces the NETX.COM net-
work shell program used in earlier versions
of NetWare.
The software actually consists of a
terminate-and-stay resident (TSR) manager
(VLM.EXE) and several Virtual Loadable
Modules (VLMs), which can be loaded at
startup or as needed. The software also


Prev Page 301 Next

DOS Requester
289
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
includes modules for dealing with security,
DOS redirection, transport-layer protocols,
and NDS (NetWare Directory Services) or
bindery commands. The figure "Structure
of NetWare's DOS Requester" illustrates
the components.
VLM.EXE is the VLM manager, and
is responsible for loading the appropriate
module at the appropriate time. VLM also
controls memory usage and communication
between relevant modules.
CONN.VLM is the Connection Table
Manager, which allows clients to connect to
a network (assuming, at least for now, that
the user is authorized to do so).
The DOS Requester's components fit into
a three-layer structure:
I The DOS Redirector, the REDIR.VLM
module, resides at the DOS Redirec-
tion Layer. This module provides DOS
file services and callouts. This is the
topmost of the three layers.
I The Service Protocol Layer has mod-
ules for providing NetWare-specific
services, and also file, print, and secu-
rity services. The components that
make up this layer are described below.
I The Transport Protocol Layer is
the lowest of the three layers, and
is responsible for making sure packets
are transmitted and that the connec-
tion is maintained. The TRAN.NLM
module is the Transport protocol
multiplexor, and is responsible for
enabling communications between the
available protocols (IPX or TCP) and
the resources at the service protocol
layer. The IPX and TCP protocols
are handled by IPXNCP.NLM or
TCPNCP.NLM, respectively. If neces-
sary, the AUTO.VLM module can be
used to reconnect a workstation to a
server automatically-for example,
to reestablish a broken connection.
AUTO.VLM will automatically recon-
figure the system to its original state.
The following services are provided at the
Service Protocol Layer:
I NetWare services are provided to han-
dle the different flavors of NetWare:
NetWare 2.x and 3.x (which use bind-
eries), NetWare 4.x (which uses Net-
Ware Directory Services, or NDS), and
Personal NetWare. These flavors are
handled, respectively, by BIND.VLM
(for 2.x and 3.x), NDS.VLM (for 4.x),
and PNW.VLM (for Personal Net-
Ware). The module for the appropriate
protocols is determined and called by
NWP.VLM-the NetWare Protocol
multiplexor.
I
File services are handled by the
FIO.VLM (file input/output) module.
This module uses a basic file transfer
protocol by default. If desirable or
necessary, however, FIO can use spe-
cial methods when reading or writing.
These measures include using a cache
(CACHE) or a packet-burst protocol
(PBODI), or transmitting large internet
packets (LIP).


Prev Page 302 Next

290
DOS Requester
I Print services are provided by
the PRINT.VLM module. Since
PRINT.VLM uses the FIO capabili-
ties, it can use any of the special
measures listed for FIO.VLM. The
print module's behavior depends
on the settings it finds in the NET.CFG
configuration file.
I Security services (both encryption and
authentication) are provided through
RSA.NLM, a module that implements
the Rivest, Shamir, and Adleman
public-key encryption algorithm.
Unlike the NetWare shell, the DOS
Requester may be called by DOS to do a
task that is network-based and that DOS is,
therefore, unable to perform. For example,
DOS may use the DOS Requester to access
file services on a remote machine.
The DOS Requester still processes
NetWare requests to get them into the
appropriate format and then sends the
requests on to the server.
BROADER CATEGOR Y
Network Shell
COMPARE
NETX
STRUCTURE OF NETWARE'S DOS REQUESTER
AUTO
VLM. EXE
C
O
N
N
RSA NDS BIND PNW
NWP
REDIR
PRINT
FIO
TRAN
IPXNCP
TCPNCP
SECURITY
NMR


Prev Page 303 Next

Downsizing
291
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
M
Dotted Decimal
Dotted decimal, also known as dotted digit,
is the notation system used to represent the
four-byte IP (Internet Protocol) addresses.
An address in this format is called a dot
address.
SEE
IP Address
M
Double Buffering
The use of two buffers for input and output
in order to improve performance and
increase throughput. In a double-buffered
environment, one buffer is processed while
the other is filling.
M
DOV (Data Over Voice)
In communications, a strategy for transmit-
ting data over the voice channel at the same
time as a voice transmission. A human lis-
tener would not hear the data being trans-
mitted. DOV requires special equipment.
COMPARE
DUV (Data Under Voice)
M
Downgrading
In the CCITT X.400 Message Handling
System (MHS), the process of converting a
message from the 1988 MHS version format
to a format suitable for an MHS based on
the 1984 version of X.400.
M
Downlink
In telecommunications, a communications
link between a satellite and one or more
earth stations.
M
Download
To transfer data, such as a file, from a host
computer to a remote machine. For exam-
ple, the host may be a mainframe or a BBS
(bulletin board system) computer. Down-
loading requires a communications protocol
that both the host and recipient can under-
stand and use.
COMPARE
Upload
M
Downsizing
Downsizing refers to the redesign of
mainframe-based business applications
to create applications capable of running
on smaller, less expensive systems, often
local-area networks (LANs) of PCs. A
client/server architecture is the model most
often implemented during downsizing.
In moving applications from large com-
puter systems to PCs, it is possible that secu-
rity, integrity, and overall control will be
compromised. Development and training
costs for the new system can be high. How-
ever, a collection of appropriately configured
PCs, networked together, can provide more
than ten times the power for the same cost


Prev Page 304 Next

292
Downtime
as a mainframe computer supporting remote
terminals.
A more accurate term might be rightsiz-
ing, to match the application requirements
of the corporation to the capabilities of the
hardware and software systems available.
MDowntime
A machine or other device that is not func-
tioning is said to be down. Downtime is a
period during which a computer or other
device is not functioning. This is in contrast
to uptime, during which the machine is
functioning.
Note that uptime and downtime are
not synonymous with availability and
unavailability. A device may be unavailable
during uptime (for example, because of
heavy activity).
M
DP (Draft Proposal)
For some standards committees, a prelimi-
nary version of specifications or standards.
The DP is circulated for a limited time, dur-
ing which comments and critiques are col-
lected by the standards committee.
M
DPA (Demand Protocol Architecture)
In Microsoft's LAN Manager network oper-
ating system, DPA is a feature that makes it
possible to load and unload protocol stacks
dynamically. This capability makes it possi-
ble to support other network environments,
such as VINES or NetWare, in the same
machine.
DPA was originally added by 3Com to its
implementation of LAN Manager, but it has
since been added to versions supported by
other vendors.
BROADER CATEGOR Y
LAN Manager
MDPC (Deferred Procedure Call)
In Windows NT and NT Advanced Server, a
called function whose task is less important
than the currently executing function. As a
result, execution of the called function is
deferred until higher priority tasks are
completed.
M
DPMI (DOS Protected Mode
Interface)
DPMI is an interface specification from
Microsoft. The interface is designed to pro-
vide DOS extension. By providing this capa-
bility, DPMI enables DOS programs to run
in protected mode, so that they can make
use of extended memory, take advantage
of system safeguards afforded in protected
mode, and so on.
The data and execution safeguards pro-
vided in protected mode allow most pro-
grams to run as DOS tasks on their own
or under Windows 3.x. DPMI provides
enhanced capabilities for 80286 and higher
processors.
DPMI was developed partly in response
to the older VCPI (Virtual Control Program
Interface). DPMI and VCPI are incompati-
ble, so these two interfaces should not be
mixed on a network.
COMPARE
VCPI (Virtual Control Program-Interface)
SEE ALSO
DOS Extender; Protected Mode


Prev Page 305 Next

DQDB (Distributed Queue Dual Bus)
293
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
M
DQDB (Distributed Queue Dual Bus)
DQDB is a network architecture that
has been recommended by the IEEE 802.6
committee for use in metropolitan-area net-
works (MANs). DQDB has the following
characteristics:
I Operates at the bottom two layers of
the OSI Reference Model: the physical
and data-link layers. Actually, DQDB
operates at the physical layer and at
the media-access-control (MAC) sub-
layer, as defined by the IEEE 802.2
committee.
I Uses two buses for the network. Each
bus operates in a single direction, and
the buses operate in opposite direc-
tions. A node on the network may
transmit and receive on one or both
buses, depending on where the node
is located in relation to the bus ends.
I Generally uses fiber-optic cable as the
physical medium. Copper cable is gen-
erally not used, because it has difficulty
supporting both the distances and the
bandwidth that may be required for a
MAN. This may change, however, as
higher-grade copper cable becomes
available. (Note that copper cable is
used in many MANs, but as access
cable to connect individual nodes or
subnetworks to the MAN bus.)
I Can support circuit-switched voice,
data, and video, and can handle
synchronous or asynchronous
transmissions.
I Provides connection-oriented, connec-
tionless, and isochronous communica-
tions services.
I Allocates bandwidth dynamically,
using time slots.
I
Supports transmission speeds of at
least 50 megabits per second (Mbps),
and will eventually support speeds of
about 600 Mbps.
I Uses 53-octet slots for transmissions.
The performance of a DQDB configura-
tion is independent of the number of nodes
and of the distances involved, which makes
DQDB ideal for high-speed transmissions.
DQDB uses a dual-bus topology, with the
buses transmitting in opposite directions.
The first node in each direction is the head
of the bus. This node has special responsibil-
ities for the bus, including the task of gener-
ating the slots in which data are transmitted.
Since the head node is at the starting end
of the bus, all other nodes on the bus are
down the line, or to move the metaphor
(and the bus) to the water, downstream from
the head node. Conversely, the head node is
up the line or upstream from all the other
nodes on the bus. Node positioning is
important when controlling access to the
network.
The DQDB architecture may use either
the "traditional" open bus topology shown
in the figure "DQDB with open bus topol-
ogy," or the looped bus shown in the figure
"DQDB with looped bus topology." Because
the looped bus topology is easier to reconfig-
ure if a node goes down, it is used more
DQDBTopology


Prev Page 306 Next

294
DQDB (Distributed Queue Dual Bus)
DQDB WITH OPEN BUS TOPOLOGY
DQDB WITH LOOPED BUS TOPOLOGY


Prev Page 307 Next

DQDB (Distributed Queue Dual Bus)
295
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
commonly. In fact, when a looped bus is
reconfigured to compensate for a lost node,
the result is an open bus.
In a looped bus topology, the head node
is also the endpoint, or tail, for the bus.
While this looks just like a ring topology,
the looped bus differs because the head node
does not pass on a transmission it receives as
the tail. Note also that the same node serves
as the head for both buses on a looped bus.
The DQDB architecture is described in
terms of three layers in the 802.6 specifica-
tions, as illustrated in the figure "Layers in
the DQDB architecture."
The DQDB layers are as follows:
Physical layer: The lowest layer, which
supports several transmission schemes.
At its lower end, this layer interfaces
to the physical medium; at the upper
end, the layer uses a convergence func-
tion to get data from the upper layer
and to prepare the data for transmis-
sion across the medium.
DQDB layer: The workhorse layer of the
DQDB architecture. It corresponds to
the lower half of the OSI Reference
Model's data-link layer, or the MAC
sublayer as specified by the IEEE 802.2
committee. The DQDB layer can pro-
vide services for any of several types of
connections. This layer is divided into
three sublayers (described later in this
article).
Outside layer: The third "layer" is not
really part of the DQDB architecture,
nor is the layer's name official. This
level is included in the specifications
in order to specify the services that
the DQDB layer must be able to pro-
vide. The description of required ser-
vices is quite heterogeneous, largely
because the DQDB architecture sup-
ports such a variety of connections
and transmissions.
To accommodate the requirements of
the layers above it, three types of services
have been defined for the DQDB layer in
the 802.6 specifications: connectionless,
connection-oriented, and isochronous.
The connectionless services do not estab-
lish a fixed connection before transmitting
data. Instead, individual packets are sent
independently of each other, possibly by dif-
ferent paths. This type of service might be
requested by the LLC sublayer, which makes
up the upper half of the data-link layer. The
MAC convergence function (MCF) does the
translation and preparation needed to have
the data passed down into the proper form
for transmission.
The connection-oriented services estab-
lish a connection first, then send the data
and, finally, break the connection. Because
a fixed (if temporary) connection is estab-
lished, all the data takes the same path. This
makes both the sender's and the receiver's
jobs a bit easier.
The isochronous services assume a con-
stant transmission pace. Such transmissions
are often synchronous, but this is not
required.
DQDB Structure
DQDB Layer Services


Prev Page 308 Next

296
DQDB (Distributed Queue Dual Bus)
LAYERS IN THE DQDB ARCHITECTURE


Prev Page 309 Next

DQDB (Distributed Queue Dual Bus)
297
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
The DQDB layer is divided into three
sublayers:
I The topmost layer interacts with the
"outside" layer; that is, it interacts
with the applications that want (or
need) to use the DQDB. At this layer,
functions are specified and/or defined
for the three main types of services
(connectionless, connection-oriented,
and isochronous) provided by the
DQDB layer.
I The middle layer provides functions
for arbitrating access to the network.
Two types of slots are used: queued
arbitrated (QA) and prearbitrated
(PA). The QA slots carry asynchronous
data from either connectionless or
connection-oriented services. The PA
slots carry isochronous data.
I
The bottom sublayer provides access
to the physical medium for both asyn-
chronous and isochronous data. This
sublayer also includes functions for
controlling the configuration and
for serving as the head of the bus.
An MCF is defined for the top DQDB
sublayer. This function does the prepara-
tions for data using connectionless services.
A connection-oriented convergence func-
tion (COCF) has been proposed in the 802.6
documents, but has not yet been defined.
Similarly, the function needed for handling
isochronous data has been proposed but not
defined.
DQDB Sublayers
Information moves around a DQDB net-
work in 53-octet slots, and slots from differ-
ent nodes are intermingled in the network
traffic. This means that nodes need to be
able to break higher-layer packets into 52-
byte chunks before sending the information.
Nodes also must be able to reconstruct a
packet from the slots received in a transmis-
sion. The 52 bytes will contain pieces of a
higher-level packet. The fifty-third byte in
a slot is for access control information.
The head node is responsible for creating
empty slots and sending these down the line,
where the slots will be used by nodes to send
their messages. By generating as many slots
as needed, the head node can make sure that
each node on the bus gets access.
To do this, the head node must know
how many slots are needed by the nodes.
Suppose a node (N) wants to transmit on
one of the buses (let's say bus A). In order to
get a slot on bus A, N must indicate-on bus
B-that N needs a slot. This request will
eventually reach the head node for bus A,
which will increment a counter that indi-
cates the number of slots A needs to create.
Bus A creates empty slots and sends these
down the line. As the slots move down the
line, they are taken by the nodes that have
requested them. These nodes fill the slots
and send them toward their destination. A
node will take only the slot it has requested,
even if that node needs additional slots since
its last request.
There are restrictions built into the slot
request and generation process to help
ensure that the slots are being allocated
DQDB Operation


Prev Page 310 Next

298
Draft Proposal (DP)
fairly and that the architecture's bandwidth
is being allocated in a balanced fashion.
MDraft Proposal (DP)
For some standards committees, a prelimi-
nary version of specifications or standards.
The DP is circulated for a limited time,
during which comments and critiques are
collected by the standards committee.
MDRAM (Dynamic Random-Access
Memory)
DRAM is a type of chip memory in which
information is stored in capacitors, whose
charge must be refreshed periodically. This
is in contrast to SRAM (static random-
access memory) in which information is
stored differently.
Dynamic RAM is slower but much
cheaper than SRAM and is, therefore, much
more widely used. Most of the chip memory
in a PC (stand-alone machine or network-
based workstation) is DRAM. If SRAM
chips are used at all, they may be used
for cache storage.
BROADER CATEGOR Y
Memory
COMPARE
SRAM (Static Random-Access Memory)
MDRDA (Distributed Relational Data
Architecture)
A distributed database architecture from
IBM. DRDA forms the core of the data-
base management capabilities in IBM's
SystemView network management package.
M
Drive
A drive is a data storage location. Drives
may be the following:
I Physical, such as floppy disk drives,
hard disk drives, or tape drives.
I Logical, such as hard disk partitions
or NetWare drives. Logical drives
represent organizational entities.
I Virtual, such as RAM disks or virtual
disks. These use physical resources to
mimic physical drives, but their con-
tents disappear when the computer is
turned off.
In the DOS environment, drives are refer-
enced by letters. For example, A: and B:
represent floppy disk drives on a PC. In a
NetWare network, drives A: through E: rep-
resent local drives on a workstation; drives
F:, G:, and so on, are logical network drives.
SEE ALSO
Directory; Drive Mapping
MDrive Mapping
The process of assigning a hard disk
volume or directories on this volume
to a particular logical disk drive is called
drive mapping, or simply mapping. For
example, a workstation user might use
drive mapping to designate the server's
hard disk as logical drive H: (from the
workstation's perspective).
Each user can have his or her own set
of drive mappings, which can be loaded
into the user's working environment when
logging on to the network or specified dur-
ing regular operation.


Prev Page 311 Next

Driver
299
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
In NetWare and other operating systems, MDriver
it is possible to map a drive letter to a partic-
ular directory on the server. In effect, this
mapping makes the directory the root of the
specified drive. Drive mapping gives a user
immediate access to the directory, and is one
way of dealing with path name restrictions
(as discussed in the Directory article).
NetWare supports four types of drive
mappings:
I Local mappings, which are to local
hard disks and floppy drives. By
default, drives A: through E: may
be used for local mappings.
I Network mappings, which are to vol-
umes and directories on the network.
By default, drives F: through Z: may
be used for network mappings.
I Network search mappings, which are
to directories that contain programs or
data files. Users can specify conditions
and rules under which search directo-
ries will be checked. See the Search
Drives article for more information.
I Directory map objects mappings,
which allow a Directory map object
to reference the location of commonly
used files or applications.
Drive mappings can be temporary or per-
manent in NetWare. Temporary mappings
disappear when a session is ended.
SEE ALSO
Search Drives
A program that serves as an interface
between two programs or between a pro-
gram and a hardware component. For
example, to ensure that a network inter-
face card (NIC) will work with a network
software package, drivers are used.
In Windows NT and NT Advanced
Server (NTAS), the term driver is used more
broadly, and also encompasses file systems,
such as the file allocation table (FAT) used
by DOS and the high performance file sys-
tem (HPFS) used by OS/2.
Drivers can be written for virtually any
kind of device or interface, including the
following:
I Printers, scanners, disks, monitors,
and other devices
I SCSI, RS-232, RS-422, IDE, and other
interfaces
I NICs, such as for Ethernet and
Token Ring
Drivers are often specialized; a particular
driver may support a single device model for
a particular program. However, rather than
creating drivers for every model, manufac-
turers may create a more or less generic
driver interface, and then encourage devel-
opers to adapt the interface for their prod-
ucts to this generic interface. Vendors may
also adapt generic drivers to handle the
special features of particular products.
Types of Drivers


Prev Page 312 Next

300
Drop
In local-area networking, two generic driver
interfaces are widely supported:
I NDIS (Network Driver Interface
Specification), developed jointly by
Microsoft and 3Com for LAN Man-
ager, but now used for other network
packages as well.
I ODI (Open Data-link Interface), an
alternative to NDIS developed by Nov-
ell for its NetWare products. It is cur-
rently less widely used than NDIS, but
is nonetheless widely supported.
Both of these represent efforts to provide
a general interface between NICs and the
higher-level protocols supported in a partic-
ular network.
NDIS and ODI provide generic interfaces,
but specific drivers for particular adapters
are also still used, partly because specific
drivers can optimize the performance of the
product. Most adapters ship with dozens of
drivers.
SEE ALSO
NDIS (Network Driver Interface Specifi-
cation); ODI (Open Data-link Interface)
M
Drop
An attachment to a horizontal cabling sys-
tem (for example, through a wallplate). This
is generally the point through which a com-
puter or other device is connected to the
transmission medium on a network. A drop
is also known as a drop line.
MDrop Box
In an AppleShare server, a term for a folder
for which write (Make Changes) but not
read privileges are granted. Users can add
items to the folder but cannot open the
folder or see its contents.
M
Dropout
Temporary loss of the signal in a transmis-
sion, such as through malfunction, power
loss, or interference.
M
Drop Set
All the components needed to connect a
machine or other component to the horizon-
tal cabling. At a minimum, this includes
cable and an adapter or connector.
M
Drop Side
All the components needed to connect a
machine or other component to the patch
panel or punch-down block that connects
to the distribution frame.
UPDATING DRIVERS
Because the driver program is generally a small
piece of software, it's relatively easy to change.
For this reason, drivers tend to be updated fairly
frequently. Vendors can generally tell you
whether their drivers have been updated, and
several magazines list driver updates as a regular
feature.
Keep your drivers up to date, but make sure you
can return to an older driver-in case incompati-
bilities develop with the newer version.
NIC Driver Interfaces


Prev Page 313 Next

DS (Directory Service)
301
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
M
DS (Digital Service)
DS is a communications service that uses
digital signaling methods. More specifically,
DS represents a telecommunications service
in North America, which defines a four-level
transmission hierarchy, with increasing
bandwidths.
DS uses pulse code modulation (PCM) to
encode an analog signal in digital form. The
signal is sampled 8000 times per second, and
each sample value is encoded in an 8-bit
value. The signal transmission uses time
division multiplexing (TDM).
DSx, (Digital Signal, where x is 0, 1, 1C, 2,
3, or 4) represents a hierarchy of channel
capacities for digital signals. The hierarchy
defines protocols, framing format, and even
the signal frequency used at the specified
level.
The DS in DS0, DS1, and so on, is some-
times expanded to digital service. The terms
are sometimes written as DS-0, DS-1, and
so on.
The data signals are transmitted over
T-carrier lines, such as T1 or T3. The higher-
capacity channels are based on the 64 kilobit
per second (kbps) DS0 channel. The DS0
channel is based on the 4 kilohertz (kHz)
analog channel used for ordinary voice
communications.
The 1.544 megabit per second (Mbps)
DS1 channel is constructed of the smaller
DS0 channels. Twenty-four DS0 channels
are multiplexed into a single DS1 channel,
yielding a 1.536 Mbps bandwidth for data.
An extra framing bit is added to each 192-
bit (eight bits per channel 24 channels)
frame. This is known as the 193rd bit, and
it represents the extra 8 kbps in the DS1
channel capacity.
Either of two techniques is commonly
used to handle framing in DS1 channels: D4
or ESF. The signals in a DS1 channel can be
transmitted over T1 lines.
Lower-capacity digital channels are also
possible. These channels are also built up
by combining DS0 channels, which can be
transmitted over fractional T1 (FT1) lines.
An FT1 line consists of one or more DS0
channels.
Higher-capacity channels are built by
multiplexing lower-bandwidth channels,
together with framing and administrative
overhead. The overhead bits are transmitted
in separate channels, which may have 8, 16,
or even 64 kbps bandwidths.
The figure "Digital signal hierarchy for
North America" summarizes the digital
signal hierarchy as it is defined in North
America. The channel configurations are
somewhat different in Europe and Asia.
To give you a sense of the relative sizes
involved in the DS hierarchy, if a DS0 chan-
nel were represented as being an inch thick,
a DS4 channel would be wider than a foot-
ball field.
SEE ALSO
D4 Framing; DACS (Digital Access and
Cross-Connect System)
M
DS (Directory Service)
Directory-related services, as defined in the
CCITT X.500 model, or naming services
as provided in Novell's NDS and Banyan's
StreetTalk. Directory services are provided
at the application layer.
DS1DS4 Levels


Prev Page 314 Next

302
DSA (Directory Service Area)
SEE ALSO
NDS (NetWare Directory Services);
StreetTalk; X.500
MDSA (Directory Service Area)
In telephony, a term used to describe the
calling area covered by a directory service.
M
DSA (Directory System Agent)
In the CCITT X.500 Directory Services
model, software that provides services for
accessing, using and, possibly, for updating
a directory information base (DIB) or tree
(DIT), generally for a single organization.
DIGITAL SIGNAL HIERARCHY FOR NOR TH AMERICA


Prev Page 315 Next

DSU/CSU (Data Service Unit/Channel Service Unit)
303
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
SEE ALSO
X.500
M
DSA (Distributed Systems
Architecture)
An OSI-compliant architecture from
Honeywell.
MDSC (Data Stream Compatibility)
In IBM's SNA (Systems Network Architec-
ture), a basic, bare-bones printing mode.
COMPARE
SCS (SNA Character String)
M
DSE (Data Switching Equipment)
Equipment used in a switching network,
such as X.25.
M
DSI (Digital Speech Interpolation)
In digital telecommunications, a strategy for
improving the efficiency of a communica-
tions channel. DSI works by transmitting
during the "quiet" periods that occur in nor-
mal conversation. DSI can nearly double the M
number of voice signals that can be carried
on the line.
M
DSOM (Distributed System Object
Model)
IBM's implementation of the CORBA (Com-
mon Object Request Broker Architecture)
object request broker from the OMG
(Object Management Group).
SEE ALSO
CORBA
M
DSP (Digital Signal Processor)
A device that can extract and process
elements from a stream of digital signals.
M
DSP (Domain Specific Part)
In the OSI Reference Model, part of the
address for the network-layer service access
point (NSAP). The DSP is the address within
the domain, which is the part of the network
under the control of a particular authority
or organization.
SEE ALSO
SAP (Service Access Point)
MDSPU (Downstream Physical Unit)
In a ring topology, a device that lies in the
direction of travel of packets.
MDSR (Data Set Ready)
A signal from a modem, sent when the
modem is ready to operate. In the RS-232C
interface, this signal is transmitted on pin 6.
DSU/CSU (Data Service Unit/Channel
Service Unit)
In digital telecommunications, the DSU and
CSU are two components of a DCE (Data-
Communications Equipment) device. These
components provide access to digital ser-
vices over DDS, T1, and other types of lines.
The DSU performs the following tasks:
I Connects to the DTE (usually a router
or remote bridge) through a synchro-
nous serial interface, which is a V.35


Prev Page 316 Next

304
DSX1/3 (Digital Signal Cross-Connect between Levels 1 and 3)
or an RS-422 connection; RS-232 con- M
nections are also possible for subrate
(low-speed) services
I Formats data for transmission over the
digital lines
I Controls data flow between the net-
work and a CSU
The CSU, which must be certified by the
FCC (Federal Communications Commis-
sion), does the following:
I Terminates the long-distance connec-
tion at the user's end
I Processes digital signals for the digital
lines
I May test remote loopback on the lines
I Serves as a buffer to keep faulty sub-
scriber equipment from bringing down
the digital service
Functionally, the DSU/CSU component
is comparable to a modem; each mediates
between a digital computing element and a
transmission medium. The medium is analog
in the case of the modem and digital for the
DSU/CSU.
The figure "DSU/CSU devices provide
access to digital lines" shows how this com-
ponent fits into a networking scheme.
MDSX1/3 (Digital Signal Cross-Connect
between Levels 1 and 3)
In digital communications, DSX1/3 specifies
the interfaces for connecting DS1 and DS3
signals (which entails connecting T-1 and
T-3 lines).
DTAM (Document Transfer and
Manipulation)
DTAM provides the communication
functions for the ITU's (International Tele-
communication Union) application-layer
Telematic services. Telematic services are
communications services other than tele-
phony and telegraphy. These include teletex
(basically, souped-up telex), fax transmis-
sion, and telewriting (transmission of hand
drawing or writing, so that the resulting
image is duplicated at the receiving end).
The DTAM specifications cover three
service classes, which specify-at a very
general level-the actions allowed on docu-
ments. The service classes are bulk transfer
(BT), document manipulation (DM), and
bulk transfer and manipulation (BTM).
Each service class is defined by more primi-
tive functional units and by communication
support functions.
To transfer documents, DTAM uses
either application level support functions or
session layer services. In the latter case-
known as transparent mode bulk transfer-
DTAM bypasses the presentation layer and
sends the material directly to the session
layer. This is allowed only in cases where the
received document just needs to be sent on
to another location. Since the recipient acts
as an intermediary, no presentation of the
document is necessary. Transparent mode is
allowed only for Group 4 faxes, which are
not yet widely used.
In normal mode, DTAM uses the services
of the ACSE (Association Control Service
Element), the RTSE (Reliable Transfer
Service Element), or the ROSE (Remote


Prev Page 317 Next

DTAM (Document Transfer and Manipulation)
305
a
b
c
d
efghijkl
m
n
o
p
q
r
s
t
u
v
w
x
y
z
Operation Service Element)-depending on
the required task.
Documents that the DTAM can handle
must conform to the ODA (Open Document
Architecture) standard. This standard is
used for the interchange of compound docu-
ments-that is, of documents that may con-
tain graphics, video, or sound in addition
to text.
The DTAM protocols provide the means
by which two DTAM service elements
(DTAM-SEs)-or rather two applications
using DTAM-communicate. The commu-
nication support functions help pass packets
(known as PDUs, or protocol data units) up
or down in the OSI hierarchical model. The
Figure "DTAM model" illustrates the hier-
archical as well as the lateral relationships.
So far, the DTAM protocol supports over
a dozen different types of PDUs. For exam-
ple, the DINQ (D-initiate request) PDU is
used for the Association Use Control func-
tional unit. This unit is the one that controls
whether there is any association between
DTAM entities at either end of the
connection.
Since several of the functional units have
yet to be finalized, there's a good chance that
more PDU types will be defined.
PRIMAR Y SOURCES
ITU recommendations T.431, T.432, and
T.433. T.62bis provides guidelines for
transmissions that bypass the presenta-
tion layer and communicate directly with
the session layer.
SEE ALSO
ACSE, ODA, ROSE, RTSE
DSU/CSU DEVICES PROVIDE ACCESS TO DIGITAL LINES


Prev Page 318 Next

306
DTAM (Document Transfer and Manipulation)
DTAM MODEL
(Session - Layer Services)
(Presentation - Layer Services)
ACSE
(Association Control
Service Element)
RTSE
(Reliable Transfer
Service Element)
DTAM User
DTAM Service (T. 400)
DTAM-SE (DTAM Service Element)
Document Body
Document structure (ODA)
Content structure (ODA)

Functional Units
Association use control
Capability
Document transfer
Document manipulation
Document
Interchange
format
Document
transfer,
document
manipulation, etc.
DTAM Protocol
PSAP
PSAP
PSAP
SSAP
SSAP
* PSAP :
presentation - layer
service access point
* SSAP :
session - layer
service access point


Prev Page 319 Next

DTS (Digital Termination Service)
307
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
MDTE (Data Terminal Equipment)
In telecommunications, a terminal, a PC,
or another device that can communicate
with a DCE (data communications equip-
ment) device. For example, in analog tele-
communications, a modem serves as a DCE,
and provides access to the telephone lines; in
digital communications, a DSU/CSU pro-
vides access to the lines for a DTE.
MDTMF (Dual Tone Multifrequency)
DTMF is a telephone technology that makes
it possible to create 16 different tones using
eight frequencies. These 16 tones suffice to
provide a unique tone for each of the 12
base buttons on a Touch Tone telephone,
as well as for up to four additional keys.
The figure "Frequencies for buttons on
a Touch Tone telephone" shows how the
frequencies are assigned to the buttons.
MDTR (Data Terminal Ready)
In the RS-232 interface, a control signal
used to indicate that a device (for example, a
computer) is ready to send and receive data.
This signal is sent on pin 20.
MDTR (Dedicated Token Ring)
DTR is a variant of the standard token
ring technology. In DTR, a direct connection M
is possible between a node and the token
ring switch. Such a node could then make
use of the entire network bandwidth, since
there are no other nodes that can share it.
ASTRAL (Alliance for Strategic Token Ring
Advancement and Leadership) is supporting
both DTR and token ring switches. The
IEEE 802.5 committee-which is the work-
ing group for token ring topology-will
wait and see whether it proves viable and
becomes widely used before committing
to the new technology.
BROADER CATEGOR Y
Token Ring
DTS (Digital Termination Service)
In telecommunications, a service by which
private networks can get access to carrier
networks using digital microwave equip-
ment within a frequency band allocated by
FREQUENCIES FOR BUTTONS
ON A TOUCH TONE TELEPHONE


Prev Page 320 Next

308
DUA (Directory User Agent)
the FCC (Federal Communications Commis-
sion) for this purpose.
MDUA (Directory User Agent)
In the CCITT X.500 Directory Services
model, a program that provides access to
the directory services. The DUA mediates
between an end-user or a client program and
a directory system agent (DSA), which pro-
vides the requested services.
SEE ALSO
X.500
M
Dual-Attachment Concentrator
(DAC)
SEE
DAC (Dual-Attachment Concentrator)
M
Dual-Attachment Station (DAS)
SEE
DAS (Dual-Attachment Station)
M
Dual Cable System
A broadband wiring arrangement in which
separate cables are used for transmission
and receiving. Such a wiring system may be
used, for example, in a 10Broad36 broad-
band Ethernet or a broadband (IEEE 802.4)
token-bus architecture.
COMPARE
Split Cable System
M
Dual Homing
In networking, a configuration in which
a node can be connected to the network
through more than one physical link. If one
link fails, the station can still communicate
via the other link.
M
Duty Cycle
In an electrical signal, the proportion of a
time period during which the signal is on,
which is when it represents a bit value of 1.
MDUV (Data Under Voice)
In telecommunications, a strategy for trans-
mitting voice and data over the same line.
COMPARE
DOV (Data Over Voice)
M
Dynamic Addressing
In an AppleTalk network, dynamic address-
ing refers to a strategy by which nodes auto-
matically pick unique addresses. A new node
keeps trying addresses until it finds one that
is not already claimed by another node.
Dynamic addressing is also referred to as
dynamic node addressing.
Dynamic addressing works as follows:
I The node selects a valid address at
random and sends an enquiry control
packet to that address.
I If the address belongs to a node, the
node responds with an acknowledge
control packet. The new node then
selects another address at random
and repeats the process.
I If the address does not belong to a
node, the enquiring node takes it as
the node's new address.


Prev Page 321 Next

Dynamic Routing
309
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
rstuvwxyz
BROADER CATEGOR Y
AppleTalk
M
Dynamic Configuration
In networking, a system capability in
which the file server can allocate memory
as needed, subject to availability, while
the network is running. Dynamic reconfig-
uration enables the server to allocate more
resources (such as buffers, tables, and so on)
as necessary in order to avoid congestion or
overload on the network.
M
Dynamic Routing
In various networking environments,
automatic rerouting of data transmissions
in order to maximize throughput or to
balance traffic on transmission channels.
Routing decisions are based on available
and acquired data about network traffic
patterns. Dynamic routing is also known
as dynamic adaptive routing.


Prev Page 322 Next

310
Dynamic Routing


Prev Page 323 Next



Prev Page 324 Next

312
Dynamic Routing


Prev Page 325 Next

EE


Prev Page 326 Next

312
E1 Carrier
EM
E1 Carrier
In digital telecommunications, E1 is a carrier
channel configuration defined by the CCITT
and used in Europe, Mexico, and South
America. Like the T carrier channels (T1,
T2, and so on) defined in North America,
the E1 carrier channel is built up of 64 kilo-
bit per second (kbps) voice channels. See the
DS (Digital Service) article for a discussion
of how the T-carrier channels are defined.
The E1 carrier is defined as thirty 64 kbps
voice channels and two 64 kbps signaling
channels. In ISDN B and D channel termi-
nology, this type of carrier is known as
30B+2D. The E1 carrier has a bandwidth of
2.048 megabits per second (Mbps).
E1 links can be multiplexed into higher-
capacity carriers. The figure "Hierarchy of
E1-based digital carriers" shows the E1 car-
rier hierarchy, which is analogous to the T1
hierarchy defined for digital communica-
tions in North America, Australia, and
Japan. Because the hierarchy also allocates
channels for link management and signaling,
the data rates are higher than the number of
64 kbps channels indicates.
BROADER CATEGOR Y
Digital Communication
COMPARE
T1 Carrier
M
EARN (European Academic and
Research Network)
A European network that provides file trans-
fer and e-mail (electronic mail) services for
universities and research institutions.
M
Earth Station
The ground-based portion of a satellite com-
munications system is called an earth station
or a ground station. The station consists of
an antenna and receiver (or transceiver) that
are in communication with a satellite in geo-
synchronous orbit.
Signals can be beamed from an earth sta-
tion to the satellite and from there to the
destination node (another earth station).
These communications services can be
leased from various companies. For long dis-
tances, the prices are competitive with earth-
based connections (such as leased or public
lines).
The size of the antenna required to
receive signals at an earth station depends
on the transmission frequency. For 19.2
kilobit per second (kbps) lines, an antenna
of about 1.2 to 3 meters (4 to 10 feet) in
diameter is sufficient. For faster speeds (such
as the 1.544 megabit per second speed of T1
lines), larger antennas are required. These
are harder to install and maintain, and may
require special permits.
M
EBCDIC (Extended Binary Coded
Decimal Interchange Code)
EBCDIC (pronounced "eb-se-dic") is an 8-
bit character encoding scheme used on IBM
mainframes and minicomputers. Compare it
with ASCII, which is used on PCs.
MECB (Electronic Cookbook)
An operating mode for the Data Encryption
Standard.


Prev Page 327 Next

ECC (Error Correction Code)
313
a
b
c
d
efghijklmnopqrst
u
v
w
x
y
z
SEE
DES (Data Encryption Standard)
M
ECC (Error Correction Code)
In digital communications, a term applied
(sometimes incorrectly) to any of several
types of codes used to detect or correct
errors that may arise during transmission.
SEE
Error Detection and Correction
HIERARCHY OF E1-BASED DIGITAL CARRIERS


Prev Page 328 Next

314
Echo
M
Echo
As a verb, echo refers to the display of typed
text on the screen. Other definitions discuss
the term in particular contexts, such as elec-
trical signaling.
M
Echo, Electrical
In electrical transmissions, an echo is a sig-
nal that "bounces off" the destination sta-
tion (or an intermediate station) and is
reflected back toward its source. The echo is
a weaker version of the original signal, and
it will interfere with any incoming signal,
which can lead to noise and transmission
errors.
An echo can occur if the transmission
lines are not properly terminated or if there
is an electrical mismatch (for example, in
impedance levels) between the sending and
receiving stations.
To eliminate the disruptive effect of an
echo, a device called an echo canceler can be
used. This device makes a copy of the echo
and superimposes a displaced copy on the
echo in order to cancel the echo signal and
remove it from the transmission lines.
An echo suppressor can also be used to
eliminate echo signals. An echo suppressor
does the same thing as an echo canceler, but
works differently.
MEcho/Echo Reply
In networking environments, echo signals
can be used to determine whether target
nodes are able to receive and acknowl-
edge transmissions. The echo signal is
sent out, and the sender waits for an
acknowledgment.
The method provides a simple mechanism
for checking network connections. With this
scheme, a node sends an Echo packet to a
destination to determine whether the desti-
nation is connected. If the destination is con-
nected and able to communicate, it responds
with an Echo Reply packet.
This echoing strategy is quick and dirty,
but only minimally informative. Further-
more, packet delivery may be unreliable
because most Echo/Echo Reply schemes are
transmitted at the network layer, which may
not guarantee packet delivery. One way to
increase reliability is to repeat the echo sig-
nal a number of times to test the connection.
The proportion of trials that are success-
ful will shed light on the reliability of the
connection.
The error-signal strategy for simple net-
work monitoring is used in several network
protocols, including ICMP (Internet Control
Message Protocol), AppleTalk, XNS (Xerox
Network Services), and Novell's IPX (Inter-
net Packet Exchange).
Most network management packages use
more powerful protocols, such as SNMP or
CMIS/CMIP for monitoring network activ-
ity. See the SNMP (Simple Network Man-
agement Protocol) and CMIS (Common
Management Information Services) articles
for more information about these protocols.
M
ECL (Emitter-Coupled Logic)
A logic scheme for very high-speed digital
circuitry. Compare ECL with CMOS (com-
plementary metal-oxide semiconductor)
and TTL (transistor-transistor logic).


Prev Page 329 Next

ECN (Explicit Congestion Notification)
315
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
rstuvwxyz
M
ECMA (European Computer
Manufacturers Association)
An association that provides technical com-
mittees for other standards organizations,
such as the ISO and CCITT.
M
ECN (Explicit Congestion
Notification)
In frame-relay transmissions, ECN is a
mechanism for indicating that there is traffic
congestion on the network. Such congestion
can be indicated in either or both of two bit
values in a packet header:
I The BECN (Backward Explicit Con-
gestion Notification) bit is set in frame-
relay headers moving in the direction
opposite the congestion and serves to
warn source nodes that congestion is
occurring "down the line."
I The FECN (Forward Explicit Conges-
tion Notification) bit is set in frame-
relay headers to warn a destination
node that there is congestion.
The figure "Use of ECN bits to signal
congestion" shows how these bits are used
for signaling if there is congestion around
node B.
BROADER CATEGOR Y
Frame Relay
USE OF ECN BITS TO SIGNAL CONGESTION


Prev Page 330 Next

316
ECNE (Enterprise Certified NetWare Engineer)
M
ECNE (Enterprise Certified NetWare
Engineer)
A title given to people who have successfully
met the requirements for CNE (Certified
NetWare Engineer) and who pass several
additional courses and tests in order to be
able to troubleshoot and operate enterprise-
wide networks.
In addition to being a CNE, ECNE
candidates must demonstrate mastery of
advanced concepts related to the NetWare
operating system (either version 3.11 or
4.x-depending on the candidate's special-
ization) and of topics selected from various
electives. Elective areas include such topics
as Internetworking products, UnixWare, and
NetWare programming.
SEE ALSO
CNA; CNE; CNI
MED (End Delimiter)
A field in a token ring token or data frame.
ED indicates the end of a token or data
frame.
SEE
Token Ring
M
EDI (Electronic Data Interchange)
EDI provides specifications for business
transactions that are done electronically-
for example, on a network. EDI standards
specify the type of information that needs to
be available or exchanged for various types
of transactions. The standards also specify
the format this information must have.
EDI services can translate data into the
appropriate formats and can send and
receive such formats. EDI services and stan-
dards support multiple protocols and multi-
ple platforms. For example, EDI services
may run on mainframes, minicomputers,
or PCSs; the services may run under VMS,
MVS, UNIX, Windows, and so on. Data
can be transmitted using various protocols,
including the ITU's (International Telecom-
munication Union, formerly the CCITT)
X.400 message handling systems.
EDI activities are broken down into
transaction sets and functional groups.
A transaction set consists of data that is
exchanged between parties to produce an
interchange (of forms, funds, etc.). For
example, the transmission of a purchase
order, an insurance form, or an invoice can
all be transaction sets. A functional group
consists of several similar transaction sets
(such as five invoices).
The transaction set is made up of seg-
ments. Each segment is either an administra-
tive chunk (such as a header or trailer) or
part of the data being exchanged (for exam-
ple, an invoice, purchase order, or other type
of form). With certain exceptions, segments
are transmitted in a predefined sequence,
and some segments may be repeated. Each
segment in a transaction set is either manda-
tory, optional, or floating. Allowable data
segments are defined and described in the
Data Segment Dictionary.
Data segments are, in turn, made up
of data elements. A data element is the
smallest unit of information in EDI. The
allowable data elements are described in
EDI Services


Prev Page 331 Next

EDI (Electronic Data Interchange)
317
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
the Data Element Dictionary. In their re-
spective dictionaries, each data segment and
data element are assigned unique identifica-
tion numbers, and each will have one or
more attributes and values associated with
it. When reading articles about EDI, it's not
uncommon to find references to particular
forms or items by number.
The corpus of documents, forms, and
other items in the world of EDI is enormous.
This is so, in part, because standards have
been developed for entire industries (trans-
portation, health care, finance, etc.). Some
of these industries are known for their
bureaucratic excesses, and one of the goals
of EDI is to help save time, work, paper, and
money by automating much of the work
and by maintaining records in electronic (as
opposed to paper) form as much as possible.
Various surveys and studies have found
that companies can save anywhere from a
few percent to almost 90 percent on relevant
transactions by switching to EDI. It's not
uncommon for a company to report savings
of $10 or more on each purchase order, for
example.
Note that for such savings to be realized,
both parties involved in a transaction must
use EDI. In fact, one reason EDI continues
to grow is that companies who are using
EDI may require prospective suppliers or
partners to use EDI in their dealings with
the company. Once these suppliers have
switched, they may, in turn, require that
their clients use EDI.
In the United States, most of the work on
EDI specifications and standards has been
done by ANSI X12 committees-actually,
by subcommittees that address more specific
topics. Over two dozen task and work
groups from the various subcommittee areas
have met or are meeting. For example, an
Interactive EDI work group and a Data
Security task group have been formed by the
X12C subcommittee, which is concerned
with communication and controls. That is,
the X12C subcommittee is concerned with
making sure information can move
smoothly, quickly, and securely over elec-
tronic lines. Other subcommittees include:
X12E (product data), X12F (Finance),
X12G (Government with, surprisingly, just
two task groups), X12I (Transportation),
and X12N (Insurance with, not surprisingly,
a dozen Work Groups and ten task groups).
Other standards for EDI also exist. For
example, continental Europe uses ODETTE
(Organization for Data Exchange by Tele-
Transmission in Europe) and the United
Kingdom uses TRADACOMS (Trading
Data Communications Standards).
Internationally, the ISO's (International
Standardization Organization) EDIFACT
(EDI for Administration, Commerce, and
Transport) standard is considered the official
specification. This is sometimes also known
as the UN/EDIFACT standard, where UN
represents the United Nations.
The various national standards organiza-
tions all have the option of being repre-
sented in the EDIFACT committees-either
directly or through another organization.
For example, the United States and several
South American countries comprise the
PAEB (Pan American EDIFACT Board).
Members of the PAEB represent US interests
in EDIFACT-at least in part.
EDI Standards and Variants


Prev Page 332 Next

318
EDO (Extended Data Out)
The X12 and the EDIFACT specifications
are not identical, and there is some contro-
versy as to whether the United States will
accept the ISO version as official. Currently,
the agreement is that by 1997, there will be
only a single EDI standard.
Eventually, EDI is expected to make up a
large part of the traffic in X.400 systems-
possibly as e-mail traffic-and also in CTI
(computer-telephone integration) systems.
PRIMAR Y SOURCES
ISO recommendation 9735; various
ANSI documents including X12.3 (Data
Element Dictionary) and X12.22
(Data Segment Dictionary); CCITT
recommendation x.435
M
EDO (Extended Data Out)
A variant of dynamic random access
memory (DRAM) that helps improve mem-
ory speed and performance. By altering the
timing and sequence of signals that activate
the circuitry for accessing memory locations,
EDO keeps data in currently accessed loca-
tions available even while beginning the next
memory access. Not all processor chip sets
support EDO RAM.
MEEMA (European Electronic Mail
Association)
A European association of developers and
vendors of electronic mail products. The
EMA (Electronic Mail Association) is the
counterpart in the United States.
M
EFF (Electronic Frontier Foundation)
The EFF is an organization founded in 1990
to help ensure that the "electronic frontier"
remains accessible and open to everyone.
The EFF tries to accomplish its goals by
providing a forum for the discussion of
issues related to the use of electronic net-
works, and a voice for end-users in public
policy and other debates.
On occasion, EFF also provides a legal-
defense fund for Sysops and other computer-
using individuals being prosecuted by the
government.
M
Effective Bandwidth
The central part of the total bandwidth in a
communications channel. This is the section
in which the signal is strongest and clearest.
The effective bandwidth is generally the area
within which the total attenuation is less
than 3 decibels (dB). (A 3 dB attenuation
corresponds roughly to a 50 percent reduc-
tion in signal strength.)
M
Effective Isotropic Radiated Power
(EIRP)
SEE
EIRP (Effective Isotropic Radiated
Power)
GETTING IN TOUCH WITH EFF
To contact the EFF, write, phone, fax, or modem:
Electronic Frontier Foundation
1001 G Street NW, Suite 950
East Washington, DC 20001
Telephone: (202) 347-5400 (voice)
E-mail: eff@eff.org


Prev Page 333 Next

EFS (Error Free Second)
319
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
M
Effective Rights
In Novell's NetWare environment, effective
rights refer to the rights a user can exercise
in a particular directory or file (versions 2.x
and later) or in the Directory tree created by
the NetWare Directory Services (NDS, in
version 4.x).
Effective rights are defined with respect
to the following:
Directory rights in the file system:
Directory effective rights are deter-
mined by any trustee assignments. If
no such assignments exist, the effective
rights of a directory are determined
by the user's effective rights in the
parent directory and the directory's
Inherited Rights Mask (in NetWare
3.x) or Maximum Rights Mask
(NetWare 2.x).
File rights in the file system: File effective
rights are determined by any trustee
assignments for the file. Otherwise, the
user's effective rights in the directory
apply.
Object rights in the NDS: Object effective
rights (in NetWare 4.x only) define
what a user is allowed to do with an
object entry in the NDS Directory tree.
These rights apply to the object as a
single structure in the tree, not to the
properties associated with the object
or to the object itself. For example, if a
user has a Browse right for an object,
the user does not automatically have
access to property information.
Property rights in the NDS: Property
effective rights (in NetWare 4.x only)
define what kind of access a user has
to the information associated with
an object.
Effective rights for NDS objects and
properties are determined by:
I
Inherited rights associated with the
object or property, taking into account
any Inherited Rights Filters (IRFs) that
apply.
I Trustee assignments associated with
a user or group
I Applicable security restrictions
BROADER CATEGOR Y
NetWare
M
Effective Throughput
The number of data bits transmitted within
a given time (such as a second). This is in
contrast to ordinary, or simple, throughput,
which represents the total number of bits
(both data and administrative) transmitted.
M
EFS (End Frame Sequence)
The last field in a token ring data packet.
SEE
Token Ring
M
EFS (Error Free Second)
One second of transmission without errors.
The total or average number of EFS can be
used as an index of transmission quality.


Prev Page 334 Next

320
EIA (Electronic Industries Association)
M
EIA (Electronic Industries Association)
An association that represents American
manufacturers in standards organizations.
The EIA has published several widely used
standards, such as RS-232C, EIA-232D,
RS-422, and RS-449. These standards
govern the electrical characteristics of con-
nections between computers and other elec-
tronic devices (such as modems or printers).
The CCITT has created international ver-
sions of several EIA standards.
Reports that are concerned more directly
with communications are produced jointly
with the TIA (Telecommunications Industry
Association). For example, EIA/TIA-568
defines five categories for unshielded
twisted-pair (UTP) cable and specifies the
minimal performance requirements for each
category.
M
EIB (Enterprise Information Base)
In enterprise networks, the information base
containing management- and performance-
related information about the network.
The information in this type of database is
used by network management or monitoring
software.
MEIRP (Effective Isotropic Radiated
Power)
The strength of a signal received at an earth
station in a satellite communications system;
that is, the strength of the satellite's signal by
the time it reaches the ground. This value is
generally measured in decibels (dB).
M
EISA (Extended Industry Standard
Architecture)
EISA is an architecture for the PC expansion
bus that provides 32-bit bus access but
remains compatible with the 8- and 16-
bit ISA (Industry Standard Architecture)
that characterizes the IBM-PC and its
descendants.
This architecture was developed by a
consortium of hardware manufacturers in
response to the 32-bit proprietary Micro-
Channel architecture developed by IBM.
BROADER CATEGOR Y
Data Bus
COMPARE
ISA; MicroChannel; PCI; VESA
MEKTS (Electronic Key Telephone
System)
In telephony, a key telephone system (KTS)
that uses electrical switches. By shrinking
the entire KTS down to electronic circuitry,
it becomes easier to add features and to
install the KTS in a telephone.
SEE ALSO
KTS
M
Electrical Signal
Electrical energy (voltage or current)
transmitted as a waveform. Signals are
distinguished by their amplitude (strength),
frequency or period (repetition rate), and
phase (timing).
Communication occurs when a modulat-
ing signal (which represents information)


Prev Page 335 Next

ELS (Entry Level System) NetWare
321
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
is superimposed on a fixed carrier signal
(which serves as a baseline) and is then
transmitted. The information is represented
by changing one or more of the modulating
signal's distinguishing features.
M
Electromagnetic Interference (EMI)
SEE
EMI (Electromagnetic Interference)
M
Electronic Cookbook (ECB)
SEE
ECB (Electronic Cookbook)
M
Electronic Key Telephone System
(EKTS)
SEE
EKTS (Electronic Key Telephone System);
KTS
MElectronic Mail Association (EMA)
SEE
EMA (Electronic Mail Association)
M
Electronic Mailbox
In an e-mail (electronic mail) system, a direc-
tory provided to store messages for a single
user. Each e-mail user has a unique ID and
a unique mailbox.
SEE ALSO
E-Mail
MElectronic Switching
In circuit switching, hardware in which the
connections are made electronically (rather
than electromechanically).
M
Elevator Seeking
Elevator seeking is a technique for optimiz-
ing the movement of the read/write heads in
a file server's hard disk.
Requests for disk access from different
nodes are queued on the basis of the heads'
position; that is, requests for data from the
same area of the disk are fulfilled together.
The heads move in a sweeping motion from
the outside of the disk to the inside. This
strategy reduces read/write head activity and
greatly increases the throughput.
The name elevator seeking comes from
the fact that people going to a particular
floor get off together, regardless of when
each person got on the elevator. Similarly,
the elevator stops at floors as they are
reached, not in the order in which the floors
were requested.
M
ELS (Entry Level System) NetWare
ELS NetWare refers to low-end NetWare
products that support a limited number of
stations and a limited range of hardware.
ELS NetWare comes in two configurations:
I ELS Level I supports up to four nodes,
a few different network interface
cards, and a limited set of operating
environments.


Prev Page 336 Next

322
EMA (Electronic Mail Association)
I ELS Level II supports up to eight nodes
and a much broader range of hardware
and operating environments.
ELS products are no longer sold.
MEMA (Electronic Mail Association)
An association of developers and vendors of
electronic mail products.
MEMA (Enterprise Management
Architecture)
EMA is a network management model from
Digital Equipment Corporation (DEC).
With this model, DEC hopes to provide the
tools needed to manage enterprise networks,
regardless of the configurations that make
up the network. The architecture is designed
to conform to the ISO's CMIP (Common
Management Information Protocol).
The DEC Management Control Center
(DECmcc) Director implements the current
version of the EMA model. This product is
extended by several add-on products that
are designed for specialized management
tasks.
In order to achieve vendor and protocol
independence, the EMA isolates the Director
as much as possible from implementation
details. The Director is in charge of manag-
ing network elements, and it uses several
kinds of modules for its tasks:
I Access modules, to provide a path to
the network elements being managed.
Each access module supports a single
type of network element, such as a
bridge or a device belonging to a par-
ticular type of network. Access mod-
ules use widely supported protocols,
such as the CMIP and the Internet
community's SNMP, to communicate.
I Functional modules, to provide
the capabilities for carrying out the
performance, configuration, security,
and other types of management tasks.
I Presentation modules, which provide
an integrated, standardized interface
for the Director.
The other major component of the EMA
model is the Executive. This element con-
tains the information about the network
elements in a Management Information
Repository.
BROADER CATEGOR Y
Network Management
ME-Mail (Electronic Mail)
E-mail (also written as email) is an applica-
tion that provides a message transfer and
storage service for the nodes on a network
or internetwork or for a stand-alone
machine through a dial-up service. Each user
has an electronic mailbox (a unique direc-
tory for storing electronic mail), and other
users can send e-mail messages to the user at
this mailbox.
The e-mail messages are sent to an e-mail
address. For the end-user, an e-mail address
is generally written as a sequence of names,
separated by periods or other special charac-
ters, as in fiddle@faddle.edu.
Once the message is stored in the recipi-
ent's mailbox, the owner of the mailbox can
retrieve whatever messages look important
and/or interesting. E-mail packages differ in


Prev Page 337 Next

E-Mail (Electronic Mail)
323
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
the ease with which such selections can be
made and also in the services the packages
provide.
All e-mail packages will send and deliver
mail, and all can let users know when they
have mail. Most packages allow you to
create the message by using the e-mail soft-
ware or by using your own resources. Many
packages also allow recipients to reply to a
message by simply annotating the original
message. Some packages allow voice mail,
which requires additional hardware.
Setting up a proprietary e-mail service
on a single network is generally straightfor-
ward, but may be of little value in the long
run. In order to exchange e-mail with users
on other networks or in remote locations,
more powerful software is needed.
E-mail services are also available through
dial-up services such as CompuServe and
MCI Mail.
If an e-mail message cannot be delivered,
it may be stored temporarily in a post office.
This is just a service with available storage
and with the ability to check periodically
whether the recipient is ready to take deliv-
ery. E-mail handling is an example of the
more general store-and-forward strategy.
The first e-mail systems were developed in
the late 1960s and early 1970s. These were
mainly small-scale, departmental systems-
although the ARPANET was a major factor
in the development of electronic messaging.
These systems were also mainly proprietary,
with little effort being made to enable e-mail
systems to communicate with each other-
even within the same company. The first
e-mail systems consisted of little more than
file transfer capabilities.
In the late 1970s and early 1980s, public
e-mail services became available through
service providers such as AT&T Mail, MCI
Mail, and CompuServe. For the most part,
mail services on these providers were used
by businesses and by individuals. Research
and academic e-mail services developed on
what was becoming the Internet.
At the same time, PCs appeared and
quickly became extremely popular. By the
mid- to late 1980s, e-mail packages for
LANs were appearing and proliferating.
As was the case with public and corporate
e-mail services, each package had its own
formats and protocols.
As mail and messaging services became
more popular and more widely used, the
need for interoperability grew. As a result,
standards were developed:
I The X.400 series of recommendations
from the CCITT (Consultative Com-
mittee for International Telegraphy
and Telephony, now going under the
name International Telecommunica-
tions Union, or ITU) provided stan-
dards for electronic messaging and
mail. The first version of the X.400
standards appeared in 1984, and these
are known as MHS 84 (for message
handling system, 1984). X.400 systems
commonly serve as a backbone for
delivering mail between (possibly
incompatible) e-mail systems.
I The SMTP (Simple Mail Transfer Pro-
tocol) in the IP (Internet protocol) suite
provided e-mail standards and proto-
cols for the Internet.
History and Overview of Electronic Mail


Prev Page 338 Next

324
E-Mail (Electronic Mail)
In the late 1980s and early 1990s, e-mail
continued to grow rapidly in popularity.
During this period, formats became more
standardized, and even the LAN-based
packages began to support either X.400
or SMTP or both.
Two other events have helped make elec-
tronic mail a truly international service:
I The appearance of the CCITT X.500
standards for directory naming and
services helped make it possible to
keep track of addresses and locations
more easily and in a more consistent
manner. During this same period, a
new version of the X.400 MHS stan-
dards appeared-known as MHS 88.
I The appearance of gateways, which
could serve as a transfer place between
incompatible mail systems-sort of
like the locks in the Panama canal pro-
vide a transfer between incompatible
oceans.
The mid- to late 1990s promise to be an
even more exciting period for electronic
mail. Several kinds of developments are
likely to take place during this period:
I Increasing bandwidth, so that even
huge files can be sent quickly and eas-
ily via e-mail. The planning and work
are already underway for gigabit-level
bandwidths for such services, and even
terabit-speed networks are beginning
to be discussed.
I Support for video, audio, and graphics
in a mail or message service. The
Multi-purpose Internet Mail Exten-
sions (MIME) provide guidelines for
how such materials should be handled.
While these represent a start, it's likely
that major developments will occur in
this area.
I
The appearance of intelligent agents to
help in mail handling and delivery, and
also to help users screen their mail.
I
The development of wireless mail ser-
vices will continue, helping to spur
advances in wireless networking.
I The generalization of electronic mail
and messaging to encompass electronic
commerce-for example, through EDI
(electronic data interchange).
I The use of e-mail as a medium for
workflow messages and traffic. Work-
flow software is used to specify or
manage the sequence of tasks needed
to carry out and complete a project-
particularly when the project requires
the participation of multiple workers.
I The use of encryption, digital signa-
tures, and other security techniques to
keep the content of e-mail messages
hidden from unauthorized eyes. This is
an essential development if e-mail is to
become a vehicle for electronic com-
merce. PEM (privacy enhanced mail) is
an example of such a security measure.
The more general PGP (pretty good
privacy) algorithm may also be used
for encryption).


Prev Page 339 Next

Emoticon
325
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
The architecture of an e-mail system can
vary, but all e-mail systems need to provide
the following types of services:
I Terminal and/or node handling, so that
the mail service can understand user
requests and respond to those requests.
I File handling, so that electronic mes-
sages can be stored as files in the
appropriate mailbox. These are gen-
eral file handling abilities, with a few
exceptions.
I Communications handling, so that a
mail server (for example) can talk to
and exchange messages with another
server at a remote site. For the most
part, these are general communications MEMI (Electromagnetic Interference)
capabilities.
I Local mail services, so that a mail
server can receive and deliver mail
from local users.
I Mail transfer, so that a mail server can
deliver electronic messages to another
server and can receive electronic mes-
sages from the other server.
Encryption and multicast capabilities are M
also common e-mail system features.
MHS (Message Handling Service) is Nov-
ell's e-mail system for NetWare. MHS is a
store-and-forward system that also provides
gateways into other messaging systems,
most notably, into X.400 systems.
Until recently, the e-mail universe was filled
with proprietary protocols, few of which
could talk to each other. Fortunately, this
has changed. Most e-mail products now
support either or both of two widely used
standards: the SMTP (Simple Mail Transfer
Protocol) from the TCP/IP protocol suite or
protocols specified in the CCITT's X.400
series of standards.
SEE ALSO
MHS (Message Handling System); MIME
(Multipurpose Internet Mail Extensions);
PEM (Privacy Enhanced Mail); PGP
(Pretty Good Privacy)
M
Embedded SCSI
A hard disk with a SCSI interface and a
controller built into the hard disk.
Random or periodic energy from external
sources that can interfere with transmissions
over copper cable. EMI sources can be
artifacts (such as motors or lighting-par-
ticularly fluorescent lighting) or natural
phenomena (such as atmospheric or solar
activity). Compare this with RFI (radio
frequency interference).
EMM (Expanded Memory Manager)
An EMM is a program that provides access
to expanded memory.
SEE ALSO
Memory Management
M
Emoticon
In electronic communication, emoticons
are special symbols that are used to convey
E-Mail System Components
E-Mail Protocols


Prev Page 340 Next

326
EMS (Expanded Memory Specification)
emotions (elation, disappointment, and so
on) or commentary (sarcasm, irony, and the
like) related to the text. Emoticons are also
known as smileys.
Emoticons are built using characters
available on any keyboard. For example, the
emoticon ;-) represents a wink, which can
convey irony, sarcasm, or a conspiratorial
"nudge-nudge, know what I mean." The
following are examples of emoticons:
:-)
Smile; happiness; agreement;
laughter
:-(
Frown; unhappiness; disagree-
ment; anger
;-)
Half-smile; irony; sarcasm;
joking
(@w@)
Amazement; incredulity
;-o
Shout
;-r
Disgust; displeasure (tongue
sticking out)
PRIMAR Y SOURCE
Smileys by David W. Sanderson (O'Reilly
& Associates) includes more than 650
symbols.
M
EMS (Expanded Memory
Specification)
In the DOS environment, the specification
for expanded memory (a type of memory
that is allocated on separate boards and
whose contents are paged into "ordinary"
memory piecemeal). Although the EMS calls
for expanded memory to have its own hard-
ware, various memory managers and drivers
can emulate expanded memory in extended
memory.
SEE ALSO
Memory
MEmulation
A complete functional duplication of one
machine or device by another. For example,
a PC may emulate a 3270 terminal in order
to communicate with an IBM mainframe. A
hardware device or a software package that
provides emulation is called an emulator.
SEE ALSO
Terminal Emulation
MEncapsulation
In a layered networking model, encapsula-
tion refers to a process by which each layer
subsumes the PDU (protocol data unit) from
the layer above into a larger PDU by adding
a header to the higher-layer PDU. (A PDU is
a packet built at a particular layer, which is
used for communicating with a program at
the same layer on a different machine). For
example, a transport-layer protocol encap-
sulates a PDU from the session layer.
The layer is often indicated by adding an
initial letter to PDU. For example, a presen-
tation layer PDU would be written as PPDU
or P-PDU.
Encapsulation is used by internetwork
links, such as certain routers or gateways.
Encapsulating routers operate at the net-
work layer, and transport-layer gateways
operate at the higher, transport layer.
The inverse process-removing the lower-
layer headers at the receiving end-is known
as decapsulation.


Prev Page 341 Next

Encoding, Signal
327
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
M
Encoding
Encoding is a process by which informa-
tion in one form or at one level of detail is
represented in a different form or at a
different level.
The term is widely used, and encoding is
practiced in many contexts. For example,
encoding may be used in the following ways:
I In text processing, characters, digits,
and other symbols are represented as
decimal values between 0 and 128
or between 0 and 255. ASCII and
EBCDIC are examples of character-
encoding schemes.
I In telegraphy, characters and digits are
represented as sequences of dots and
dashes. Morse code is an example of
this encoding scheme.
I In the transmission of digital signals
over networks, binary values (0 and 1)
are represented as changes in voltage
or current levels. Signal-encoding
schemes include AMI (Alternate Mark
Inversion), Manchester, Differential
Manchester, and MLT-3.
Special forms of encoding include transla-
tion and compression. In translation, one
encoding scheme is converted to another,
such as from EBCDIC to ASCII. In com-
pression encoding, redundant information
is represented in a more efficient manner.
In the X.400 Message Handling System
(MHS), a distinction is made between two
Encoding Contexts
Special Forms of Encoding
types of encoding for a packet: definite or
indefinite. A definite encoding scheme
includes explicit length information in a
packet. This information is generally stored
in a length field.
An indefinite encoding scheme uses a spe-
cial character (EOC, for end of content) to
indicate when the end of a packet is reached.
Note that encoding here refers to the
form a packet takes, rather than the form
an electrical signal takes.
M
Encoding, Signal
Signal encoding is a set of rules for repre-
senting the possible values for an input sig-
nal in some other form. For example, in
digital communications, the signal-encoding
rule will determine what form an electrical
signal will take to represent a 1 or a 0.
Dozens of rule sets have been proposed
just for digital signals. Each has its advan-
tages and disadvantages. In the simplest
encoding scheme, a particular voltage level
represents one value and a different (or
zero) voltage represents a different value.
For binary inputs, just two different voltage
levels are needed.
Note that the actual voltage levels and
charges used to represent the bit values
depend on the logic being used for the cir-
cuitry. TTL logic is used in situations where
circuit speed is important; because of its
lower voltage requirements, CMOS logic is
used where low power consumption is more
important (for example, in battery-powered
computers).
It is possible to encode more than one bit
in a digital signal. For example, by allowing
four different voltages, you can represent
two bits in each signal; with eight voltages,


Prev Page 342 Next

328
Encoding, Signal
you can represent three bits at a time, and
so on. The trade-off is that the components
must be able to make finer discriminations,
which makes them more expensive to manu-
facture or more error prone, or both.
Each signal has a predefined duration, so
that the voltage for a single signal will be
held for a specified amount of time. The
shorter this time needs to be, the faster the
potential transmission speed. The trade-off
is that the faster signal allows less room for
distortion by noise, so that the error rate
may increase.
In order to distinguish the individual bits
in a series of the same bit values, such as a
series of 1 values in succession, sender and
receiver may use clocking (a timing mecha-
nism used to determine the start of a bit sig-
nal) to establish the duration of a signal for
a single bit, which is called the bit interval.
Each party in the communication uses its
own clock to time the signal.
Since transmission speeds can be more
than 100 megabits per second (Mbps), the
clocks must be very closely synchronized. In
practice, the clocks may need to be resyn-
chronized millions of times per second.
To avoid the overhead of inserted clock-
ing bits, most encoding schemes use actual
bit values (generally a 1) as the clocking bit.
This works fine unless there are long
stretches without any 1 values. (At high
speeds, a "long" stretch can be as short as a
single byte.) For these cases, special, adap-
tive encoding schemes, such as B8ZS (bipo-
lar with 8 zero substitution), have been
developed to make sure such a sequence
never occurs.
Signal Timing
Some encoding schemes are self-clocking,
in that the clocking is built into the signal
itself. This clocking usually takes the form
of a voltage change at the middle of the bit
interval.
Although self-clocking schemes make
external clocks and adaptive encoding
unnecessary, they cannot operate at more
than half the speed of the system clock. This
is because two clock cycles must be used to
split a bit interval in half.
Some encoding schemes use transition cod-
ing in which a value is encoded by a transi-
tion (from one voltage level to another)
during the bit interval. For example, the rep-
resentation of a 1 in a scheme with transi-
tion coding may consist of a positive voltage
for half the bit interval and zero voltage for
the other half. This type of encoding scheme
is also self-clocking. Transition coding tends
to be less susceptible to noise.
The following general encoding schemes
summarize a few of the strategies used to
represent binary values.
Unfortunately, there is little consistency
in signal-encoding terminology, so that the
same term may refer to two different encod-
ing schemes. If the encoding method is
important for your purposes, ask the vendor
for sample timing diagrams, so that you can
see the actual encoding.
Unipolar: Uses a positive or a negative
voltage (but not both in the same
Self-Clocking Encoding Schemes
Transition Coding
A Sampling of Encoding Schemes


Prev Page 343 Next

Encoding, Signal
329
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
scheme) to represent one value (for
example, 1) and a zero voltage to rep-
resent the other. Unipolar encoding
does not use transition coding, and it
requires an external clock.
Polar: A positive voltage represents one
value and a negative voltage represents
the other. Polar encoding does not use
transition coding, and it requires an
external clock.
Bipolar: Uses positive, negative, and zero
voltages, usually with zero voltage rep-
resenting one value and a nonzero
voltage representing the other. Bipolar
encoding may use transition coding,
and it may be self-clocking.
Biphase: Includes at least one transition
per bit interval. In addition to making
this scheme self-clocking, the transi-
tion coding also makes it easier to
detect errors. Biphase schemes are
often used for networks.
Of these schemes, variants on bipolar and
biphase are the most widely used. The fol-
lowing sections describe some specific ver-
sions of bipolar and biphase strategies. In a
specific communications context, a binary
value may undergo several encoding
schemes before actually being transmitted.
AMI, also known as ABP (alternate bipolar)
encoding, is a bipolar scheme. This signal-
encoding method uses three possible values:
+V, 0V, and V (positive, zero, and negative
voltage). All 0 bits are encoded as 0V
(zero voltage); 1 bits are encoded as +V
and V (positive and negative voltage) in
alternation. The figure "AMI encoding
for a bit sequence" shows an example of
AMI encoding.
AMI encoding is used in DSx-level trans-
missions, as in ISDN (Integrated Services
Distributed Network), FDDI (Fiber Distrib-
uted Data Interface), and other high-speed
network architectures.
AMI encoding is not self-clocking. This
means that synchronous transmissions, such
as those using digital signal methods, must
use an external clock for timing. The posi-
tive and negative voltages associated with
1 bits are used for this timing.
In order to ensure that the transmission
never gets out of synch, some environments
require a minimum density of 1 values in
any transmission. The minimum pulse den-
sity is generally set to at least one in every
eight bits. To ensure that this pulse-density
requirement is met, a variant encoding
method, called B8ZS, is used.
Like AMI, B8ZS uses three possible values:
+V, 0V, and V (positive, zero, and negative
voltage). All 0 bits are encoded as 0V (zero
voltage); 1 bits are encoded as +V and
V (positive and negative voltage) in
AMI (Alternate Mark Inversion)
AMI ENCODING FOR
A BIT SEQUENCE
B8ZS (Bipolar with 8 Zero Substitution)


Prev Page 344 Next

330
Encoding, Signal
alternation. Unlike AMI, however, B8ZS
requires that at least one bit out of every
eight must be a 1; that is, eight consecutive
0 values will never occur in B8ZS.
If eight consecutive 0 bits are encoun-
tered, the encoding will insert a 1 before the
eighth 0. This value will be removed at a
later point. A minimal density of 1 values
is needed because these values are used for
timing. If the transmission contains too long
a string of 0 values, the sender and receiver
can get out of synch without knowing it. By
ensuring there will be at least one opportu-
nity to synchronize every eight bits, the
transmission can never get too far out of
synch.
Differential Manchester is a biphase signal-
encoding scheme used in Token Ring local-
area networks (LANs). The presence or
absence of a transition at the beginning of a Manchester
bit interval indicates the value; the transition
in mid-interval just provides the clocking.
For electrical signals, bit values will gen-
erally be represented by one of three possible
voltage levels: positive (+V), zero (0V), or
negative (V). Any two of these levels are
needed-for example, +V and V.
There is a transition in the middle of
each bit interval. This makes the encoding
method self-clocking and helps avoid signal
distortion due to DC signal components.
For one of the possible bit values but not
the other, there will be a transition at the
start of any given bit interval. For example,
in a particular implementation, there may be
a signal transition for a 1 bit. The figure
"Differential Manchester encoding for a bit
sequence" shows an example of a signal
Differential Manchester
using +V and V, with signal transition on
1 bits.
In differential Manchester encoding, the
presence or absence of a transition at the
beginning of the bit interval determines the
bit value. In effect, 1 bits produce vertical
signal patterns; 0 bits produce horizontal
patterns, as shown in the figure. The transi-
tion in the middle of the interval is just for
timing.
Manchester is a biphase signal-encoding
scheme used in Ethernet LANs. The direc-
tion of the transition in mid-interval (nega-
tive to positive or positive to negative)
indicates the value (1 or 0, respectively) and
provides the clocking.
The Manchester scheme follows these
rules:
I +V and V voltage levels are used.
I There is a transition from one to the
other voltage level halfway through
each bit interval.
I There may or may not be a transition
at the start of each bit interval,
depending on whether the bit value
is a 0 or 1.
DIFFERENTIAL MANCHESTER
ENCODING FOR A BIT SEQUENCE


Prev Page 345 Next

Encoding, Signal
331
a
b
c
d
e
f
g
h
ijklmn
o
p
q
r
s
t
u
v
w
x
y
z
I For a 1 bit, the transition is always
from V to +V; for a 0 bit, the transi-
tion is always from +V to V.
In Manchester encoding, the beginning of
a bit interval is used merely to set the stage.
The activity in the middle of each bit inter-
val determines the bit value: upward transi-
tion for a 1 bit, downward for a 0 bit. The
figure "Manchester encoding for a bit
sequence" shows the encoding for a sample
bit sequence.
MLT-3 is a three-level encoding scheme that
can also scramble data. This scheme is one
proposed for use in FDDI networks. An
alternative is the two-level NRZI.
The MLT-3 signal-encoding scheme uses
three voltage levels (including a zero level)
and changes levels only when a 1 occurs. It
follows these rules:
I +V, 0V, and V voltage levels are used.
I The voltage remains the same during
an entire bit interval; that is, there are
no transitions in the middle of a bit
interval.
MANCHESTER ENCODING
FOR A BIT SEQUENCE
MLT-3 Encoding
I The voltage level changes in succes-
sion: from +V to 0V to V to 0V to
+V, and so on.
I The voltage level changes only for a
1 bit.
MLT-3 is not self-clocking, so that a syn-
chronization sequence is needed to make
sure the sender and receiver are using the
same timing. The figure "MLT-3 encoding
for a bit sequence" shows an example of this
encoding.
NRZ, also known as differential encoding,
is a bipolar encoding scheme that changes
voltages between bit intervals for 1 values
but not for 0 values. This means that the
encoding changes during a transmission. For
example, 0 may be a positive voltage during
one part and a negative voltage during
another part depending on the last occur-
rence of a 1. The presence or absence of a
transition indicates a bit value, not the volt-
age level.
MLT-3 ENCODING FOR
A BIT SEQUENCE
NRZ (Non-Return to Zero)


Prev Page 346 Next

332
Encoding, Signal
NRZ is inexpensive to implement, but it
is not self-clocking. It also does not use tran-
sition coding.
The figure "NRZ encoding for a bit
sequence" shows the encoding for a sample
bit sequence.
RZ is a bipolar signal-encoding scheme that
uses transition coding to return the signal to
a zero voltage during part of each bit inter-
val. It is self-clocking.
The figure "Differential and nondifferen-
tial RZ encoding of a bit sequence" shows
both differential and nondifferential ver-
sions of the RZ encoding scheme. In the dif-
ferential version, the defining voltage (the
voltage associated with the first half of the
bit interval) changes for each 1 bit and
remains unchanged for each 0 bit.
In the nondifferential version, the defin-
ing voltage changes only when the bit value
changes, so that the same defining voltages
are always associated with 0 and 1. For
example, +5 volts may define a 1, and -5
volts may define a 0.
FM 0 (frequency modulation 0) is a signal-
encoding method used for LocalTalk net-
works in Macintosh environments. FM 0
uses +V and V voltage levels to represent
bit values. The encoding rules are as follows:
I 1 bits are encoded alternately as +V
and V, depending on the previous
NRZ ENCODING FOR
A BIT SEQUENCE
RZ (Return to Zero)
DIFFERENTIAL AND
NONDIFFERENTIAL RZ
ENCODING OF A BIT SEQUENCE
FM 0 Encoding


Prev Page 347 Next

Encryption
333
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
voltage level. The voltage level remains
constant for an entire bit interval for a
1 bit.
I 0 bits are encoded as +V or V, de-
pending on the immediately preceding
voltage level. The voltage changes to
the other value halfway through the bit
interval.
The figure "FM 0 encoding for a bit
sequence" shows the encoding for a sample
bit sequence.
FM 0 is self-clocking because the encod-
ing for a 0 bit can be used to determine the
length of a bit interval and to synchronize
the sender and receiver.
M
Encryption
Most simply, encryption is a process in
which ordinary text or numerical informa-
tion (plaintext) is converted into an unintel-
ligible form (called ciphertext, among other
terms) using a well-defined (and reversible)
conversion algorithm and a predefined bit
value (known as a key). The key provides a
starting value for the encryption algorithm.
For various reasons, some information
must be kept encrypted. Because of the
fervor with which this statement is believed,
encryption has become an active area of
research and study. Much computing and
brain power has been expended in develop-
ing encryption algorithms that are impossi-
ble to crack and then cracking them.
Three broad strategies can be used for
encryption: the traditional strategy, the
private-key strategy, and the public-key
strategy.
The traditional encryption strategy is simply
to devise and apply a conversion algorithm.
The receiver must know the algorithm and
the key in order to reverse the conversion
and decrypt the information. This approach
has two weaknesses:
I The algorithms and keys used tend, as
a class, to be easier to crack than those
used in the other strategies.
I The algorithm or key may be stolen or
intercepted while being communicated
to the receiver.
Secret-key encryption strategies use a single
key-known only to the sender and the
receiver-and a public encryption algo-
rithm. Private-key encryption is also known
as one-key key, single key, or symmetric key
encryption.
The Data Encryption Standard (DES),
which was adapted in 1977 as the official
United States encryption standard for non-
classified data, uses a secret-key strategy.
The encryption algorithm is quite complex
and involves numerous permutations and
FM 0 ENCODING FOR
A BIT SEQUENCE
Traditional Encryption
Secret-Key Encryption


Prev Page 348 Next

334
End Bracket
transpositions of message elements. See the
DES article for more information. Different
levels of encryption can be used to make the
ciphertext even more unintelligible.
As long as the secret keys are kept secret,
this encryption strategy is very effective. For
example, even though it uses only 56 bits
for the encryption key, the DES has an
extremely small likelihood of being cracked.
Secret-key strategies have one major dis-
advantage: it is not possible to protect a
message against fraud by either the sender
or the receiver.
Public-key encryption strategies use the two
halves of a very long bit sequence as the
basis for the encryption algorithm. Public-
key encryption is also known as double-key
encryption or asymmetric key encryption
One key (one half of the bit sequence)
is placed in a public-key library to which
everyone has access. The other key is known
only to a single party, and is this party's pri-
vate key. Either half of the bit sequence can
be used to encrypt the information; the other
half is needed to decrypt it. Someone wish-
ing to send a message can use the receiver's
public key to encrypt the message; the
receiver can use the private key to decrypt it.
To reverse the process, the erstwhile receiver
uses the private key to encrypt the message.
The destination party can use the public key
to decrypt the message.
This encryption strategy is simple to
implement. It is also relatively easy to crack
unless the initial bit sequence is quite long.
The RSA algorithm is an exception to this
weakness and has the advantage of being
able to protect against fraud by the sender
or receiver. See the RSA Algorithm article
for more information.
MEnd Bracket
A circuit board with slots into which other
boards can be plugged. The motherboard in
a PC is a backplane. A segmented backplane
is a backplane with two or more buses, each
with its own slots for additional boards.
M
End Node
In a network, a station that serves as a
source or a destination for a packet. An
end node should be able to communicate
through all the layers in the OSI Reference
Model or an equivalent layered model.
SEE ALSO
Node
M
End of Content (EOC)
In telecommunications, a special character
used to indicate the end of a message or
page.
M
End Office (EO)
SEE
EO (End Office)
M
End System (ES)
In the OSI Reference Model, an end system
(ES) is a network entity, such as a node,
that uses or provides network services or
resources. An end system is known as a host
in Internet terminology.
Public-Key Encryption


Prev Page 349 Next

End System (ES)
335
a
b
c
d
e
f
g
hijklmnopqrstuvwxyz
Architecturally, an end system uses all
seven layers of the OSI Reference Model.
This is in contrast to an intermediate system
(IS), or router, which uses only the bottom
three layers (the subnet layers) of the model.
The figure "Communications involving
intermediate and end systems" shows the
relationship between intermediate and end
systems.
BROADER CATEGOR Y
OSI Reference Model
COMPARE
Intermediate System (IS)
COMMUNICATIONS INVOLVING INTERMEDIATE AND END SYSTEMS


Prev Page 350 Next

336
End-to-End Routing
M
End-to-End Routing
A routing strategy in which the entire route
is determined before the message is sent.
This is in contrast to node-to-node routing,
in which the route is built step-by-step.
M
End-User
In a network, the ultimate consumer of a
networking service.
MEnhanced Parallel Port (EPP)
SEE
EPP (Enhanced Parallel Port)
MENS (Enterprise Network Services)
ENS is an extension to Banyan's VINES net- M
work operating system (NOS). ENS enables
StreetTalk to keep track of servers using
NOSs other than VINES, such as any ver-
sion of Novell's NetWare 2.x and later or
of Apple's AppleTalk.
StreetTalk is the global naming service for
VINES. A naming service keeps track of
which nodes and devices are attached to the
network and assigns a global name to each
node. The name is independent of the partic-
ular network in which the node is located
and makes it possible for a user connected to
one server to use resources attached to a dif-
ferent server, without knowing which spe-
cific server has the resources.
ENS for NetWare is a special version for
use in NetWare versions 2.2 or later. Nam-
ing services are not needed in version 4.0,
because this version provides global naming
through the NetWare Directory Services
(NDS).
ENS for NetWare includes four
components:
I Server software, which runs the dedi-
cated server that is needed to run ENS
for NetWare
I A StreetTalk agent, which runs as a
VAP (Value-Added Process, for Net-
Ware 2.x) or as an NLM (NetWare
Loadable Module, for NetWare 3.x)
I Client software, which must run on
each workstation that wants to use
ENS
I ENS utilities, which are used instead
of NetWare utilities
Enterprise Computing
A term for networks that encompasses most
or all of a company's computing resources.
In most cases, an enterprise computing net-
work will include a whole range of com-
puters, which may be running different
operating systems and belong to different
types of networks. Consequently, one of the
biggest challenges for enterprise computing
is to achieve interoperability for all its
components.
SEE ALSO
Network, Enterprise
M
Enterprise Information Base (EIB)
SEE
EIB (Enterprise Information Base)


Prev Page 351 Next

EOC (End of Content)
337
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
M
Entity
In networking models, entity refers to an
abstract device, such as a program, function,
or protocol, that implements the services for
a particular layer on a single machine. An
entity provides services for entities at the
layer above it and requests services of the
entities at the layers below it.
The term entity is also used to refer to
a device on a network, at least when that
device is running a program or providing
a service.
M
Entrance Facilities
In a premises distribution system (PDS), the
location at which the building's wiring and
the external wiring meet.
M
Entry Point
For networking hardware, the point at
which a node is connected to the network;
for software, the point at which a program,
module, or function begins executing. In
IBM's NMA, entry point refers to the soft-
ware through which an SNA-compliant
device can communicate with the network
management program.
SEE ALSO
NMA (Network Management
Architecture)
MEntry State
In a routing table for an AppleTalk network,
a value that indicates the status of a path.
Such an entry may have the value good,
suspect, or bad, depending on how recently
the path was verified as being valid.
M
Envelope
In communications or electronic mail
(e-mail) systems, envelope refers to infor-
mation that is added to a data packet in
order to make sure the packet reaches its
destination and is received correctly. This
information is generally appended as a
header (and possibly also a trailer) for the
data packet.
In relation to an electrical signal, enve-
lope is used as a term for the signal's shape,
such as sine, square, or trapezoidal.
The term enveloping refers to a process
by which multiple faxes are included in a
single transmission.
M
Envelope Delay Distortion
In an electrical signal, the amount of delay
between different frequencies. The greater
this delay, the greater the distortion.
M
EO (End Office)
In telephony, a central office, which is where
a subscriber's lines are terminated and con-
nected to other exchanges.
SEE
CO (Central Office)
M
EOC (End of Content)
In telecommunications, a special character
used to indicate the end of a message or
page.


Prev Page 352 Next

338
EPP (Enhanced Parallel Port)
M
EPP (Enhanced Parallel Port)
A parallel port with a maximum signal rate
of 16 megabits per second (Mbps). The EPP
specifications were developed jointly by Xir-
com and Zenith, and the developers plan to
produce a 64 Mbps version. The faster port
makes external LAN cards (such as those
produced by Xircom) more viable.
M
Equalization
The process by which a device's frequency-
response is made uniform over a specified
frequency range. This is done to eliminate,
or at least decrease, distortion in a signal
due to high-frequency signals being slowed
to a greater degree than lower-frequency
waves. A device that performs equalization
is called an equalizer.
MErlang
In communications, a measure of the degree
to which a communications channel is being
used to capacity. One Erlang is defined as
36 CCS (hundreds of call seconds), which
amounts to an entire hour of channel usage
at capacity.
M
Error Detection and Correction
In communications, an error is a situation in
which the received material does not match
what was sent. Errors can arise for any of
many reasons, including the following:
I Problems with the signal, such as
noise, interference, or distortion
I Protocol problems, so that sender and
receiver cannot understand each other
I Buffer overflow, such as when the
capacity of a channel or a device is
exceeded
Error correction is a term for any of sev-
eral strategies for ensuring that the receiver
ends up with the same message as the one
originally sent. To accomplish this, two steps
are necessary: detecting an error and cor-
recting it. In digital communications, errors
are at the level of individual bits, so the task
becomes one of ensuring that the bit
sequence received matches the one sent.
Various precautions and measures can be
taken to identify and possibly even correct
errors. These measures vary in how effective
they are, and all impose a transmission
penalty in the form of extra bits that must
be sent.
Detecting errors involves the identification
of an incorrect or invalid transmission ele-
ment, such as an impossible character or a
garbled (but not encrypted) message.
In general, error-detection strategies rely
on a numerical value (based on the bytes
transmitted in a packet) that is computed
and included in the packet. The receiver
computes the same type of value and com-
pares the computed result with the transmit-
ted value. Error-detection strategies differ
in the complexity of the computed value
and in their success rate.
Error-detection methods include cyclic
or longitudinal redundancy checks and
the use of parity bits. Parity bits, CRC
(cyclic redundancy check), and LRC
Error-Detection Methods


Prev Page 353 Next

Error Detection and Correction
339
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
(longitudinal redundancy check) values
are sometimes referred to as ECCs (error
correction codes), even though, strictly
speaking, they can only help detect errors.
Hamming codes, on the other hand, are true
ECCs, because they provide enough infor-
mation to determine the nature of the error
and to replace it with a correct value.
CRC is an error-detection method based
on a transformation of the bit values in a
data packet or frame. The transformation
involves multiplying the bit pattern by a
polynomial equation, whose order depends
on the number of bits allocated for the com-
puted value. The more bits, the better the
error-detection capabilities.
The sender computes a CRC value and
adds this to the data packet. The receiver
computes a CRC value based on the data
portion of the received packet and compares
the result with the transmitted CRC value.
If the two match, the receiver assumes the
packet has been received without error. Note
that a matching CRC value is no guarantee
of an error-free transmission, although it
does make it almost certain that any errors
overlooked involved more than two bits in
the packet.
The following are some of the CRC
tests that have been developed and that
are used in communications and networking
contexts:
CRC-12: A 12-bit CRC check, used with
older protocols, most notably, IBM's
BSC (Binary Synchronous Communi-
cation) protocol.
CRC-16: A 16-bit CRC check, used in
many file transfer protocols. CRC-16
can detect all single- and double-bit
errors, all errors in which an odd num-
ber of bits are erroneous, and most
error bursts (signals in which multiple
bits in succession are erroneous, for
example, because of some temporary
glitch or interference in the power
supply).
CRC-CCITT: A 16-bit CRC check,
intended as an international standard.
CRC-32: A 32-bit CRC check, used in
local-area network (LAN) protocols
because it can detect virtually all
errors.
Parity, also known as vertical redundancy
checking (VRC), is a crude error-detection
method, which is used in serial transmis-
sions. With this method, an extra bit is
added at regular locations, such as after
seven or eight data bits. The value of the
parity bit depends on the pattern of 0 and
1 values in the data byte and on the type
of parity being used.
Bits 3, 4, and 5 in the UART (universal
asynchronous receiver/transmitter) line con-
trol register (LCR) determine the parity
setting in a serial communication. The
following values are used (with bit values
displayed in the order 345):
None (000): The value of the parity bit
is ignored.
Odd (100): The parity bit is set to what-
ever value is required to ensure that
CRC (Cyclic Redundancy Check)
Parity, or Vertical Redundancy Checking (VRC)


Prev Page 354 Next

340
Error Detection and Correction
the bit pattern (including parity bit)
has an odd number of 1 values. For
example, with 1010 1101, the parity
bit would be set to 0.
Even (110): The parity bit is set to what-
ever value is required to ensure that
the bit pattern (including parity bit)
has an even number of 1 values. For
example, with 1010 1101, the parity
bit would be set to 1.
Mark (101): The parity bit is always set
to the mark value (1).
Space (111): The parity bit is always set
to the space value (0).
Another type of parity, called block parity or
longitudinal redundancy checking (LRC), is
computed for each bit place value in a block
of bytes. For example, after every eight
bytes, an additional byte is set. One of these
extra bits corresponds to each place value
for the preceding set of bytes. Block parity is
always set to even (according to ISO stan-
dard 1155), so that each block parity bit is
set to whatever value is required to give the
column of bits an even number of 1 values.
The figure "LRC and VRC parity"
shows these two types of parity in a
single transmission.
Block Parity, or Longitudinal
Redundancy Checking (LRC)
Once an error is detected, the most common
correction scheme is to request a retransmis-
sion. The retransmission may consist of
either just the erroneous material or the cor-
rected material and all the material that was
sent after the error but before the receiver
alerted the sender. Needless to say, correct-
ing errors can become expensive if there are
a lot of them.
It is possible to develop automatic error-
correction tools. For example, forward
error correction (FEC) methods enable the
receiver to correct an error without requir-
ing a retransmission. Popular FEC methods
include Hamming and HBC (Hagelberger,
Bose-Chaudhuri) coding.
To do error-correction on the fly, many
extra bits must be added to the message in
order to locate and correct errors. (Once
located, correcting a bit-level error is really
not difficult: if 0 is wrong, then 1 must be
the value). Such methods may be used in
communications in which retransmissions
are more disruptive and/or costly than
LRC AND VRC PARITY
Error-Correction Methods


Prev Page 355 Next

ESDI (Enhanced Small Device Interface)
341
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
the overhead of sending correctable
information.
M
Error Correction Code (ECC)
In digital communications, a term applied
(sometimes incorrectly) to any of several
types of codes used to detect or correct
errors that may arise during transmission.
SEE ALSO
Error Detection and Correction
M
Error Rate
A measure of erroneous transmission ele-
ments in relation to the total transmission.
This information can be conveyed in several
ways. A widely used index is the BER, which
specifies the number of erroneous bits per
million (or billion or trillion) bits.
SEE ALSO
BER (Bit Error Rate)
M
ESCON (Enterprise System
Connection Architecture)
ESCON is a fiber-optic communications
channel. IBM developed this architecture for
use as a back-end network for connecting its
ES/9000 series (or compatible) mainframes
and peripheral devices, such as controllers,
channel extenders, and storage devices.
ESCON uses either 50/125 or 62.5/125
(core/cladding diameter) multimode fiber.
The light source for ESCON is an LED
(light-emitting diode), which sends signals
at a wavelength of approximately 1,325
nanometers (nm). This wavelength is popu-
lar because of its optical properties.
ESCON uses a 4B/8B signal-encoding
scheme, in which groups of four or eight bits
are encoded as 5- or 10-bit symbols, respec-
tively. 4B/8B is more efficient than the
Manchester or differential Manchester
signal-encoding schemes used in most local-
area networks (LANs). ESCON supports
transmission speeds of up to 200 megabits
per second (Mbps).
The optical fiber runs from the main-
frame's channel controllers to a copper-
based (not optical), switched-star concentra-
tor, which IBM calls a director. Control units
for the mainframes are connected to the
director. Concentrator and mainframe can
be 2 or 3 kilometers (1 to 2 miles) apart,
depending on whether the 50 or 62.5
nanometer fiber core is used.
The director keeps channel activity down
by sending signals only to lines for which the
signals are intended, as opposed to passing
the signals on to all lines (as a passive con-
centrator would do).
BROADER CATEGORIES
Cable, Fiber-Optic; Network Architecture
M
ESDI (Enhanced Small Device
Interface)
An interface and storage format for hard
disks. ESDI can support relatively high-
capacity (up to a gigabyte or so) drives and
supports access times as low as about 20
milliseconds.
COMPARE
IDE; SCSI


Prev Page 356 Next

342
ESF (Extended Superframe Format) Framing
M
ESF (Extended Superframe Format)
Framing
In digital signaling, ESF is a method for
framing a DS1 channel. (Framing is identify-
ing the individual channels in the DS1 chan-
nel). ESF framing groups 24 (193-bit)
frames into an ESF superframe, so that
each DS1 channel consists of one ESF
superframe.
In each ESF superframe, the values in
every 193rd bit (in bits 193, 386, and so on)
are used for any of three purposes:
I Framing, as originally intended
(frames 4, 8, 12, ..., 24).
I A 4 kbps link between endpoints
(frames 1, 3, 5, ..., 23).
I A 6-bit cyclic redundancy check (CRC)
value (frames 2, 6, 10, ..., 22)
The eighth bit in every channel of frames
6, 12, 18, and 24 is used for signaling
between central offices. The signaling capa-
bilities for ESF framing are more sophisti-
cated than for D4 framing, because four
frames provide signaling for ESF, compared
with only two frames for D4. The figure
"Elements in ESF framing" illustrates this
method.
COMPARE
D4 Framing
MESN (Electronic Switched Network)
An ESN is a telecommunications service for
private networks. A private network is one
consisting of multiple PBXs (private branch
exchanges) at various locations. ESN pro-
vides automatic switching between PBXs,
so that a PBX can be called from any other
PBX in the network without the need for
a dedicated connection between the two
PBXs.
Because a private network is also known
as a tandem network, an ESN is said to pro-
vide "electronic tandem switching."
M
Establishment Controller
In an IBM environment, an establishment
controller can support multiple devices, such
as IBM or ASCII terminals or token ring
nodes, for communication with a mainframe
host. The controller communicates with the
host's front-end processor (FEP). The IBM
3174 establishment controller is an example
of this type of controller.
If local, the link between controller and
device can be over a parallel line, an ESCON
link, or through a token ring network.
Remote connections can use V.24, V.35, or
X.21 interfaces, and SNA/SDLC, X.25, or
BSC protocols.
In IBM's SNA (Systems Network Archi-
tecture) environment, an establishment
controller is a type 2.0 PU (physical unit).
BROADER CATEGOR Y
SNA (Systems Network Architecture)
SEE ALSO
Cluster Controller


Prev Page 357 Next

Ethernet
343
a
b
c
d
efghijklmno
p
q
r
s
t
u
v
w
x
y
z
M
Ethernet
Ethernet is a shared-media network architec-
ture. Its elements are the result of work by
Xerox, Intel, and Digital Equipment Corpo-
ration. Ethernet, along with variants defined
in the IEEE 802.3 standard, is currently the
most widely used architecture for local-area
networks (LANs). According to some esti-
mates, there are more than 10 million Ether-
net nodes around the world. Estimates of
Ethernet's share of the LAN configurations
range between 60 and 90 percent.
An Ethernet network has the following
characteristics:
I Operates at the two lowest layers in
the OSI Reference Model: the physical
and data-link layer.
I Uses a bus topology. Nodes are
attached to the trunk segment, which
is the main piece of cable in an Ether-
net network. (10BaseT, a variant archi-
tecture based on the IEEE 802.3
standard, can use a star topology.)
I Can operate at a speed of up to 10
megabits per second (Mbps). Several
variants operate at slower speeds, and
newer variants promise faster speeds.
ELEMENTS IN ESF FRAMING


Prev Page 358 Next

344
Ethernet
I Uses CSMA/CD, a media-access
method based on collision detection.
This access method is specified as part
of the IEEE 802.3 document.
I Broadcasts transmissions, so that each
node gets the transmission at the same
time. A broadcast strategy is necessary
for a collision detection type of media-
access method.
I Uses Manchester encoding to represent
the 0 and 1 values that make up the
physical signal. This is a self-clocking
encoding method that includes a volt-
age transition in the middle of each bit
interval. To break a bit interval into
two halves, the clock rate must be at
least twice the maximum transmission
speed, so that a 20 megahertz (MHz)
clock is required for 10 Mbps Ether-
net. (Implementations don't actually
achieve the maximum transmission
rate, so that you can get by with
slower clocks.)
I Uses 50-ohm coaxial cable. Variants
can use 50- or 75-ohm coaxial,
twisted-pair, and fiber-optic cable.
Each type of cable has its characteristic
add-ons (connectors and terminators).
I Is a baseband network, although
variants also support broadband
networks.
The figure "Context and properties of
Ethernet" summarizes this architecture.
Ethernet's roots go back to Project ALOHA
at the University of Hawaii in the 1960s.
The CSMA/CD access method was devel-
oped for the ALOHA WAN.
Ethernet version 1.0 was superseded in
1982 by Ethernet 2.0, which is currently
the official Ethernet standard. This is also
known as DIX (for Digital, Intel, Xerox)
Ethernet or Blue Book Ethernet.
Ethernet Versions
CONTEXT AND PROPER TIES
OF ETHERNET
Context
Network Architecture
Shared-Media

ARCnet

Ethernet

Token Ring
Switched-Media
Ethernet Properties
Description
Shared-media, baseband network
Topology
Bus (Ethernet 1.0 or 2.0)

Bus or Star (802.3-based Ethernet)
Access method CSMA/CD
Speed
Up to 10 Mbps
Cable
50-ohm coaxial (Ethernet 1.0 or 2.0)

50-ohm coaxial, unsheilded twisted-pair,

Fiber-optic (802.3-based Ethernet)
Frame size
46-1500 data bytes
Variants
10Base5 (thick Ethernet)

10Base2 (thin Ethernet)

10BaseT (twisted-pair Ethernet)

10BaseF (fiber-optic Ethernet)

10Broad36

100 Mbps Ethernets (proposed)


Prev Page 359 Next

Ethernet
345
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
A variant on this standard was formu-
lated by the IEEE 802.3 work group. This
variant is sometimes called Ethernet as well.
However, although Ethernet and 802.3 are
similar, there are differences in the way the
data-link layer is handled and in the format
of a packet. These differences are explained
later in this article.
Because of these differences, difficult-
ies will arise if you try to mix different
types of Ethernet on the same network.
802.3 and Ethernet 2 nodes cannot co-
exist on the same network. Fortunately,
most implementations allow you to select
which flavor of Ethernet you want to use
on the network.
Some networking environments let you
have different types of packets on the net-
work under certain conditions. For example, Physical Layer Properties
NetWare allows both 802.2 and 802.3 pack-
ets to coexist on a network. (Packet types
are discussed later in this article.)
Ethernet networks are grouped by their
broadcast method, type of cable, and physi-
cal properties.
In a baseband network, one node can broad-
cast at a time. In a broadband network, mul-
tiple nodes can broadcast at the same time.
Blue Book Ethernet operates only in base-
band mode. Ethernet 802.3-based imple-
mentations can operate in either baseband
or broadband mode.
Ethernet Groupings
Baseband versus Broadband
Ethernet networks are also categorized
according to the type of cable used. Thin
and thick Ethernet use thin and thick coaxial
cable, respectively. Twisted-pair Ethernet is
actually an 802.3 architecture that uses
unshielded twisted-pair (UTP) cable. The
following are some of the synonyms for
these Ethernet varieties:
Thick Ethernet: ThickNet, Standard
Ethernet, 10Base5
Thin Ethernet: ThinNet, CheaperNet,
10Base2
Twisted-pair Ethernet: UTP Ethernet,
10BaseT
The IEEE 802.3 working group developed a
simple notation system to characterize vari-
ous physical-layer properties of an Ethernet
network. Ethernet networks are described
using three elements related to the wiring
and the physical signal. Each description has
three elements:
Speed/Band/Length or Cable-type
as in
10Base5
The first element, Speed, specifies the
approximate maximum transmission speed,
or bandwidth, in megabits per second
(Mbps) for the network. This will be a
1, 5, 10, or 100 (for newer, experimental
networks).
The second element, Band, is either
Base or Broad, depending on whether the
Thick, Thin, and Twisted-Pair


Prev Page 360 Next

346
Ethernet
network is baseband or broadband. For
example, 10Base5 specifies a baseband
network; 10Broad36 specifies a broad-
band network.
The third element, Length or Cable-type,
usually specifies the approximate maximum
length of a network segment in hundreds
of meters. For example, 10Base5 can have
network segments of up to 500 meters
(1,650 feet). In some cases, the length value
is specified in 50-meter increments. For
example, the 1Base5 network supports net-
work segments up to 250 meters, not 500
meters.
In other cases, the third element is used to
specify cable type. For example, 10BaseT
and 10BaseF specify networks with twisted-
pair and fiber-optic cable, respectively.
The table "Types of Ethernet Networks"
summarizes the types of Ethernet networks
that have been defined in IEEE 802.3 or by
other groups. See the 10Basex, 10Broad36,
and 100BaseT articles for more details.
Although the details differ, Ethernet net-
works all use a limited number of compo-
nents, which include Ethernet network
interface cards (NICs), cables, connectors,
transceivers and receivers, hubs, punch-
down blocks, and baluns.
Each node must have an Ethernet NIC,
which provides the computer with access to
the network. An NIC converts, packetizes,
Ethernet Hardware
Ethernet NICs
and transmits data from the computer and
receives, unpacketizes, and converts data
received over the network. NICs are
architecture-specific. This means that you
cannot use an Ethernet NIC for a Token
Ring network. It also means that you may
not be able to use an 802.3 card for an
Ethernet network or vice versa.
An Ethernet and an 802.3 card can trans-
mit packets to each other, because the Ether-
net and 802.3 packets have the same general
structure. However, the variant cards cannot
read each other's packets, because certain
fields in the packets have different types
of information. Some NICs support both
Ethernet and 802.3 formats, and are there-
fore able to read and create both types of
packets. Even if the cards cannot communi-
cate directly, the networking software will
generally be able to translate.
Ethernet NICs can have any or all of the
following connectors: BNC, DIX, RJ-xx. On
NICs with multiple connectors, you will
generally need to set DIP switches or jumper
settings on the board to indicate the type of
connector you will be using.
Ethernet cards include a hardware
address on a ROM chip. This address is
assigned by the IEEE and the vendor and is
unique to that particular NIC. Part of the
address contains vendor information, and
part identifies the board itself. This address
can be used by bridges and routers to iden-
tify a particular node on a network.


Prev Page 361 Next

Ethernet
347
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
st
u
v
w
x
y
z
Blue Book Ethernet networks use coaxial
cable. Networks based on the 802.3 archi-
tecture can use coaxial, fiber-optic, or
twisted-pair cable. The cable in an Ethernet
network may have any of several functional
uses:
Ethernet Cable
I Trunk cable is used for the main net-
work segment, which is known as the
trunk segment. Nodes are attached,
directly or indirectly, to the trunk
segment.
I Drop cable is used to attach nodes
indirectly to a trunk segment in a thick
Ethernet network. This type of cable is
TYPES OF ETHERNET NETWORKS
TYPE
DESCRIPTION
10Base2
10Base5
10BaseT
1Base5
10Broad36
10BaseF
10BaseFB
10BaseFP
10BaseFL
100BaseVG
100BaseT
Thin Ethernet using thin (3/16-inch), 50-ohm coaxial cable. This is arguably the most popular
Ethernet configuration.
Thick Ethernet using thick (3/8-inch), 50-ohm coaxial cable. Although it's the cabling for Blue
Book Ethernet, this is not a very popular configuration because thick coaxial cable is difficult to
handle and install.
Twisted-pair Ethernet using UTP cable. This configuration was adopted as the 802.3i standard in
1990, and it is becoming popular because UTP is inexpensive and easy to install and work with.
The StarLAN network developed by AT&T. StarLAN uses UTP cable and a star topology, and
was defined long before the 10BaseT standard was proposed.
The only broadband network defined in the 802.3 standard. This network uses 75-ohm coaxial
cable (CATV cable).
The only network in the 802.3 standard that explicitly calls for fiber-optic cable. This type is
actually divided into three variations: 10BaseFB, 10BaseFP, and 10BaseFL.
This network uses optical fiber for the backbone, or trunk, cable. Trunk segments can be up to
2 kilometers (1.25 miles) in length.
This specifies a network that uses optical fiber and a star topology. The coupler used to distrib-
ute the signal is passive (does not regenerate the signal before distributing). As a result, such a
network needs no electronics except for those in the computer. Maximum length for a piece of
such cable is 500 meters (1,650 feet).
This specifies a network that uses optical fiber to connect a node to a hub, or concentrator.
Cable segments can be up to 2 kilometers in length.
A 100 Mbps Ethernet network developed by Hewlett-Packard and AT&T Microelectronics.
A 100 Mbps Ethernet network developed by Grand Junction Networks. This is a proposed stan-
dard of the IEEE 802.3 study group. Variants include 100BaseT4, 100BaseTX, and 100BaseFX


Prev Page 362 Next

348
Ethernet
also known as transceiver cable
(because it connects the node to
a transceiver) and as AUI cable
(because of the type of connectors
at either end of such a cable).
I Patch cable is used in 802.3 networks
to connect any of the following: two
hubs, a node from the wallplate to a
punch-down block, or a wiring hub
to a punch-down block.
See the Cable article for more informa-
tion about network cabling.
Connectors are used to connect cable seg-
ments. An Ethernet (bus) network also needs
terminators and grounded terminators,
because network segments must be properly
grounded and terminated to prevent signals
from being reflected back over the network.
The following types of connectors are
used:
I Thick Ethernet networks use N-series
connectors and terminators on the
trunk and AUI, or DIX, connectors
on the NIC.
I Thin Ethernet networks use BNC con-
nectors and terminators on the trunk
and on the NIC.
I Twisted-pair Ethernet networks use
RJ-45 connectors or variants on these.
These networks do not require sepa-
rate terminators.
Ethernet Connectors
See the Connector; Connector, AUI; Con-
nector, BNC; and Connector, RJ-xx articles
for more information.
Repeaters clean and regenerate a sig-
nal. Repeaters are used in the middle of
a stretch of cable that is so long that the
signal quality would deteriorate to an unac-
ceptable level without regeneration. Hubs
often act as repeaters.
Transceivers can transmit and receive sig-
nals. Transceivers provide the actual point at
which the node makes contact with the net-
work. Ethernet/802.3 transceivers may be
internal (on the NIC) or external, depending
on the type of Ethernet. External transceiv-
ers, which are used for thick Ethernet, are
attached to the trunk cable with an N-series
connector or with a vampire tap.
Transceivers are called MAUs (medium
attachment units) in the IEEE 802.3
document.
Hubs are wire collectors. They are used in
802.3 networks that use twisted-pair cable.
Wires from nodes in a twisted-pair Ethernet
network may be terminated at the hub.
Hubs may be internal (boards installed in a
machine) or external (stand-alone compo-
nents). These components are also known as
concentrators.
Hardware manufacturers have created
special-purpose hubs that enhance the oper-
ation of an Ethernet network or that extend
the capabilities of certain components.
Repeaters and Transceivers
Hubs


Prev Page 363 Next

Ethernet
349
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
Examples of these are enhanced hubs and
switched hubs.
Enhanced hubs for 10BaseT networks
have been enhanced with various capabili-
ties and features by different manufacturers.
These enhancements include the following:
I Network monitoring and manage-
ment capabilities.
I Nonvolatile memory to save settings
and performance information even
during a power outage.
I
Security features, such as the ability to
send a packet only to its destination
while sending a busy signal to all other
nodes. This helps increase the security
on the system by preventing a mean-
ingful message from being intercepted
by an unauthorized node.
Switched-hub technology can increase the
effective bandwidth of an Ethernet network
by allowing multiple transmissions on the
network at the same time. For this technol-
ogy to work, the network must have multi-
ple servers, and the hub must be able to
switch to any of multiple network segments.
An Ethernet switch connects a limited num-
ber of network segments. This is in contrast
to a simple bridge, which connects two seg-
ments. Each network segment communi-
cates over the switch through its own port
on the switch. Ethernet switches operate at
the data link level (level two of the OSI hier-
archy) and work in many ways like a multi-
port bridge.
Ethernet Switches
Like a multiport bridge, an Ethernet
switch can segment a larger network-for
example, to help relieve traffic congestion by
not allowing transmissions within a segment
to leave that segment.
However, Ethernet switches have some
additional features that help make them very
popular. By placing switches intelligently in
a large network, it's possible to produce
more efficient network arrangements,
thereby resulting in faster throughput. Some
switches can even provide dedicated connec-
tions between two network segments.
Kalpana developed the first Ethernet
switch just a few years ago. Since then,
switches have become extremely popular
as one solution to the increased traffic on
Ethernet networks-with faster Ethernets
being the other. Because of their popularity,
numerous vendors now supply Ethernet
switches.
Two basic classes of Ethernet switches
are available:
I Workgroup switches communicate
with only a single node on each port.
Such a switch can provide dedicated
services between segments. Because
only a single machine can communi-
cate at each port, a workgroup switch
doesn't need to check for collisions at
the port, and it only needs minimal
resources for storing addresses. Such
switches require simpler circuitry and
so are relatively inexpensive-often
less than $300 per port.
I Network, or segment, switches are
more sophisticated and more expen-
sive. Such switches support multiple


Prev Page 364 Next

350
Ethernet
nodes at each port-and must, there-
fore, be able to store all the addresses
and forwarding information. Network
switches use the spanning tree algo-
rithm to prevent redundant paths
between segments.
A punch-down block may be used in a
twisted-pair network to provide a more
convenient location to terminate wires from
nodes in such a network. A punch-down
block is a device for making physical contact
with the wire inside a cable jacket, thereby
establishing the necessary connection for
electrical activity. Using such an intermedi-
ate connection makes it easier to change the
wiring scheme.
Baluns are used to connect coaxial cable seg-
ments (for example, an AUI cable attached
to a node) and twisted-pair cable segments
(for example, a cable attached to a hub).
Ethernet uses a bus configuration. Ethernet
802.3 networks can also use a star topology. Ethernet Operation
In a bus, nodes are attached to the net-
work's backbone, or trunk segment. Nodes
are attached directly in thin Ethernet and
with a drop cable in thick Ethernet. The fig-
ure "A thick Ethernet (bus) layout" shows
an example of a layout of a bus network.
The number of nodes that can be
attached to a trunk segment depends on the
type of cabling: a 10Base5 (thick coaxial)
segment can support up to 100 nodes; a
Punch-Down Block
Baluns
Ethernet Layout
10Base2 (thin coaxial) segment can support
no more than 30 nodes.
A link segment connects two repeaters. A
link segment is not treated as trunk segment.
You cannot attach a node to link segment
cable; you must attach the node to the trunk
segment.
Both ends of each Ethernet trunk cable
segment need to be terminated, and one of
these ends need to be grounded. Depend-
ing on the type of cable, N-series or BNC
terminators are used. If there are repeaters
connecting trunk segments, each of the seg-
ments must be terminated separately at the
repeater.
A fiber-optic inter-repeater link (FOIRL)
uses special transceivers and fiber-optic cable
for a link segment. With an FOIRL link, the
segment between the transceivers can be
up to 2 kilometers (1.25 miles).
In a star topology, such as in twisted-
pair Ethernet, the nodes are attached to a
central hub rather than to a backbone cable.
The hub serves to broadcast transmissions
to the nodes and to any other hubs attached.
The figure "Layout of a twisted-pair (star)
Ethernet network" shows the layout for a
simple star network.
An Ethernet network works as follows:
I Access: A node that wants to send a
message listens for a signal on the net-
work. If another node is transmitting,
the node waits a randomly determined
amount of time before trying again to
access the network.


Prev Page 365 Next

Ethernet
351
a
b
c
d
efghijklm
n
o
p
q
rstuvwxyz
A THICK ETHERNET (BUS) LAYOUT



Prev Page 366 Next

352
Ethernet
I Transmission: If there is no activity on
the network, the node starts transmit-
ting and then listens for a collision. A
collision occurs if another node also
found the network idle and started
transmitting at the same time. The two
packets collide, and garbled fragments
are transmitted across the network.
I
Collision handling: If there is a colli-
sion, the first node to notice sends spe-
cial jam packets to inform other nodes
of the collision. The colliding nodes
LAYOUT OF A TWISTED-PAIR (STAR) ETHERNET NETWORK
Workstation
Workstation
Workstation
Coaxial or Fiber-Optic Cable
Workstation
Coaxial or Fiber-Optic Cable
Workstation
UTP Cable
UTP Cable
UTP Cable
UTP Cable
UTP Cable
UTP Cable
MAU
MAU
File Server
Hub
The gray area around the hub indicates that the connections to the hub may not be
direct. A node or MAU may be connected directly to a wallplate, from there to a punch-
down panel, and from there to the hub.



Prev Page 367 Next

Ethernet
353
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
both retreat and wait a random
amount of time before trying again
to access the network.
I Reception: If there is no collision, the
frame is broadcast onto the network.
All nodes listen to each packet trans-
mitted. Each node checks the packet's
destination address to determine
whether the packet was intended for
that node. If so, the node processes the
packet and takes whatever action is
appropriate. If the node is not the
recipient for the packet, the node
ignores the packet. (This eavesdrop-
ping feature of Ethernet networks-
actually, of bus topologies in general-
makes it difficult to implement
message-level security on an
Ethernet network.)
Ethernet frames, or packets, come in several
flavors. However, all Ethernet frames consist
of preamble, header, data, and trailer
components.
Each of the Ethernet frame elements has
a predefined structure:
Preamble (8 bytes): Consists of eight
bytes, which are divided into seven
preamble bytes and one start frame
delimiter (SFD) byte for certain packet
flavors. These bytes are used to mark
the start of a packet and to enable the
sender and receiver to synchronize.
Header (14 bytes): Consists of three
fields: a 6-byte destination address, a
6-byte source address, and a 2-byte
field whose value is interpreted as a
length for some packet flavors and as
information about the network-level
protocol for other flavors. Interpreting
this third field as length or type distin-
guishes the two main types of Ethernet
packets (Ethernet 2 and 802.3-based
packets).
Data (461,500 bytes): Contains what-
ever packet was passed by the higher-
level protocol. Ethernet 2 packets con-
tain network-layer packets in the data
component; 802.3-based packets get
the data component from a sublayer
that may add to the network-layer
packet. The data component must be
at least 46 bytes, so it may include
padding bytes.
Trailer (4 bytes): Consists of a frame
check sequence (FCS). These bytes
represent a CRC (cyclic redundancy
SQE SUPPORT
All Ethernet variants except version 1.0 expect a
SQE (signal quality error) signal from transceiv-
ers. This signal, which is also known as a heart-
beat, "proves" that the component is working and
is, therefore, capable of detecting collisions.
Mixing components that do and don't support
SQE on the same network is asking for trouble. If
a component sends an SQE signal to a compo-
nent (such as NIC) that doesn't support SQE, the
receiver may assume the signal indicates a colli-
sion and will send a jam signal (the signal used to
stop transmission when a collision occurs).
Ethernet Frames
Ethernet Frame Elements


Prev Page 368 Next

354
Ethernet
check) value, which provides informa-
tion for detecting errors in a transmis-
sion. This component is the same in all
packet flavors.
Not counting the preamble, the three
remaining components yield Ethernet pack-
ets that are between 64 and 1,518 bytes.
The major distinction in packets is between
Ethernet 2 and 802.3-based flavors. This
distinction depends on how the values in the
third header field are interpreted. The Ether-
net packet flavors include Ethernet 2, 802.3,
802.2, and Ethernet SNAP.
File servers for Ethernet networks will
generally be able to handle multiple frame
flavors, although you may need to run a util-
ity to take advantage of this capability. With
a multi-flavor server, nodes that use different
Ethernet versions may be able to communi-
cate with each other, but only through the
server. For example, nodes using 802.3 and
Ethernet 2 NICs may be able to pass pack-
ets, but they will not be able to communi-
cate directly with each other.
Ethernet 2: This is the simplest of the
packet flavors. The third header field is
Type, and its value specifies the source
of the network layer protocol being
used. The table "Selected Ethernet
Type Field Values" lists some of the
possible values for this field. The data
component is whatever was received
by the data-link layer from the net-
work layer above it. (The other packet
formats receive the data component
from a data-link sublayer.)
802.3: This flavor has Length as the third
header field. The field's value specifies
the number of bytes in the data com-
ponent. The 802.3 flavor is sometimes
known as 802.3 raw, because it does
not include LLC (logical-link control)
sublayer information in the data com-
ponent (as does, for example, an 802.2
frame).
802.2: This packet is similar to the 802.3
format in that it has a Length (rather
than a Type) header field, but differs
in that part of the data component is
actually header information from the
LLC sublayer defined above the MAC
sublayer in the IEEE 802.2 standard.
The first three or four bytes of an
802.2 packet's data component
contain information of relevance to
the LLC sublayer. The first two bytes
contain values for the DSAP (Destina-
tion Service Access Point) and SSAP
(Source Service Access Point). These
values identify the protocols being
used at the network level.
The third byte is the Control field,
which contains information regarding
the type of transmission (such as con-
nectionless or connection-oriented)
being used. The packet passed by the
network layer follows after these three
values.
Ethernet_SNAP (Sub-Network Access
Protocol): This variant of an 802.2
packet contains LLC sublayer informa-
tion as well as five additional bytes
of information as part of the data com-
ponent. Two of the five bytes specify
the type of protocol being used at
the network layer. This is the same
Ethernet Packet Flavors


Prev Page 369 Next

Ethernet
355
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
SELECTED ETHERNET TYPE FIELD VALUES
VALUE
(HEXADECIMAL)
SOURCE
VALUE
(HEXADECIMAL)
SOURCE
0x0600
0x0800
0x0801
0x0805
0x0806
0x0807
0x0a00
0x0bad
0x6003
0x6004
0x6005
0x6010
0x7030
0x8008
0x8035
0x8038
0x803d
0x803f
0x8046
0x8065
0x809b
0x809f
Xerox XNS IDP
IP (Internet Protocol)
X.75 Internet
X.25 Level 3
ARP (Address Resolution
Protocol)
XNS Compatibility
Xerox 802.3 PUP
Banyan Systems
DEC DECnet Phase IV
DEC LAT
DEC DECnet diagnostics
3Com Corporation
Proteon
AT&T
Reverse ARP
DEC LANBridge
DEC Ethernet CSMA/CD
Encryption Protocol
DEC LAN Traffic monitor
AT&T
University of
Massachusetts, Amherst
EtherTalk (AppleTalk
running on Ethernet)
Spider Systems Ltd.
0x80c0
0x80d5
0x80e0
0x80f3
0x80f7
0x8137
0x9000
0x9001
0x9002
Digital Communications
Associates (DCA)
IBM SNA Services over
Ethernet
Allen-Bradley
AARP (AppleTalk ARP)
Apollo Computer
Novell NetWare IPX/SPX
Loopback (Configuration
test protocol)
Bridge Communications
XNS Systems Management
Bridge Communications
TCP/IP Systems Management


Prev Page 370 Next

356
Ethernet
information as in the Type field for an
Ethernet 2 packet, except that the field
is in a different location in the packet.
This Ethernet Type field is preceded by
a three-byte Organization Code field,
which specifies the organization that
assigned the Ethernet Type field value.
The table "Selected Ethernet Type Field
Values" shows a list of selected Ethernet
Type field values.
The figure "Structure of an Ethernet
frame" shows the components of the differ-
ent flavors of Ethernet frames.
A destination node checks for several types
of errors that can creep into Ethernet pack-
ets (or frames). In particular, the node
checks for each of the following types of
invalid packets:
I Long (oversized) packets are longer
than the allowed size (1,518 bytes for
Ethernet) but have a valid CRC value.
These may be caused by a faulty LAN
driver.
I Runt (undersized) packets are shorter
than the minimum size (64 bytes), but
have a valid CRC value. These may
be caused by a faulty LAN driver.
I Jabber packets are longer than 1,518
bytes and have an invalid CRC
value. These may be caused by a
faulty transceiver.
I Alignment errors are packets that have
extra bits, which means that they do
not end on byte-boundaries. Such
packets will also have invalid CRC val-
ues. These may be caused by a faulty
component (NIC, transceiver, or
cable).
I
CRC errors are packets that have a
valid number of bytes and end on a
byte-boundary but have an invalid
CRC value. These may be caused by
noise on the cable or because a cable
segment was too long.
I
Valid packets are packets that have
none of the preceding problems.
Only valid packets are passed on
the higher-level protocols in a trans-
mission. Valid packets are created
by properly functioning networking
software and hardware.
The IEEE 802.3 working group, whose task
was to formulate a standard for CSMA/CD-
based networks, came up with something
that looks like Blue Book Ethernet, but that
differs in several important ways. The Ether-
net 802.3 standard was adopted in 1985,
and the addition (802.3i) was adopted in
1990. The table "Differences between Ether-
net 802.3 and Blue Book Ethernet" summa-
rizes the distinctions between these variants.
Because 802.3 distinguishes between the
LLC and MAC sublayers, the process of cre-
ating a packet for transmission goes through
an extra level of handling. In 802.3 net-
working, a network-layer packet becomes
the data for a PDU (protocol data unit)
at the LLC sublayer. A PDU, in turn,
becomes the data when an MAC sublayer
packet is constructed for transmission
over the physical connection. In Blue Book
Ethernet networking, the network-layer
Invalid Frames
802.3 Differences


Prev Page 371 Next

357
Ethernet
STRUCTURE OF AN ETHERNET FRAME
10101010
10101010 10101011
Preamble/SFD
DA
SA
Length
FCS
8
6
6
2
4
46-1500
Preamble/SFD
DA
SA
Type
Data
FCS
8
6
6
2
4
46-1500
DSAP
SSAP
Control
1
1
1 or 2
IEEE 802.3-Based
IEEE 802.3-Based Frame Format
802.2 Frame
SNAP Frame
Ethernet 1.0/2.0
Data (from higher-level protocol)
43-1497 or 42-1496
Data (from higher-level protocol)
38-1492 or 37-1491
DSAP
SSAP
Control
Type
LLC Layer Information
LLC Layer Information SNAP Information
46-1500
Data
1
1
3
2
1 or 2
Organization Code
Preamble: 7 identical bytes; used for synchronization
SFD (Start Frame Delimiter): Indicates the frame is
about to begin
DA (Destination Address): Contains the address of the
frame's destination
SA (Source Address): Contains the address of the frame's
sender
DSAP (Destination Service Access Point): Specifies
the process receiving the packet at the destination's network layer
SSAP (Source Service Access Point): Specifies the
process sending the packet from the source's network layer
Control: Specifies the type of LLC service requested
Length: Indicates the number of data bytes (IEEE 802.3-based
variants)
Type: Indicates the upper-level protocol that is using the packet
(Ethernet 1.0/2.0 variants)
Data: Contains the information being transmitted, which may consist of
a higher-layer packet (may be padded)
FCS: A frame check sequence
Organization Code: Specifies the organization that assigned the
following Type field
Type: Indicates the upper-level protocol that is using the packet
Data


Prev Page 372 Next

358
Ethernet
packet becomes the data portion of a packet.
The figure "Layers involved in handling
Blue Book and 802.3 Ethernet packets"
illustrates the process.
A 10BaseT, or twisted-pair Ethernet, net-
work uses unshielded twisted-pair (UTP)
DIFFERENCES BETWEEN
ETHERNET 802.3 AND
BLUE BOOK ETHERNET
802.3
ETHERNET
Supports bus or star
topologies.
Supports baseband or
broadband networks.
Defines only the MAC
sublayer of the data-
link layer. Uses the
LLC sublayer defined
in the IEEE 802.2 stan-
dard for the rest of the
data-link layer.
Uses 7 bytes for a pre-
amble and 1 byte as a
start of frame delim-
iter (SFD) for a packet.
Uses the third header
field to indicate the
length of the frame's
data component.
Can use the SQE signal Can use the SQE signal
as a network manage-
ment device.
Supports only a bus
topology.
Supports only baseband
networks.
Does not divide the
data-link layer into
sublayers.
Uses 8 bytes for a pre-
amble; does not distin-
guish a separate SFD
byte.
Uses the third header
field to specify the type
of higher-layer protocol
using the data-link
services.
as a network manage-
ment device only in
version 2.0.
Twisted-Pair Ethernet
cable and a star topology, as opposed to the
coaxial cable and bus topology of Blue Book
Ethernet. In this architecture, each node is
connected to a central wiring hub, which
serves as the relay station for the network.
This 802.3-based variant was officially
adopted as IEEE standard 802.3i in 1990.
A twisted-pair Ethernet network needs
the following components:
I NIC with on-board MAU (or trans-
ceiver), to mediate between the node
and the network (one per node)
I External MAU, for mediating between
the network and nodes that use coaxial
or fiber-optic cable (optional)
I UTP cable, to connect nodes to a
wiring hub
I Wiring hubs (stand-alone or peer)
I Punch-down block, to make wire
termination more flexible and easier
to change (optional)
I RJ-45 connectors, for connecting to
wall plates and to NICs
In order to be sufficiently free of interfer-
ence, UTP cable for a network should have
enough twists in the wire. Some telephone
cable may not be suitable, because it is too
flat and has too few twists. The cable also
must have enough conductors for the eight-
wire RJ-45 connectors.
Each node in a 10-BaseT network is con-
nected directly or indirectly to a wiring hub.
Indirect connections can be through wall
plates or by connecting the PC to an exter-
nal MAU, which is connected to a wall plate
or to a hub.


Prev Page 373 Next

Ethernet
359
a
b
c
d
efghijklm
n
o
p
q
r
s
t
u
v
w
x
y
z
10BaseT networks can use either of two
kinds of hubs:
I A stand-alone hub is an external com-
ponent with RJ-45 connections to link
the nodes. This type of hub has its own
power supply.
I A peer hub is a card that can be
installed in one of the machines on the
network. This internal hub must be
connected physically to the NIC in the
machine, and it depends on the PC for
its power.
Nodes are connected to one of these
hubs-from a distance no greater than 100
meters (330 feet)-using UTP cable with
RJ-45 connections at each end. A 10BaseT
network can have up to four linked hubs.
A 10 Base5, or thick Ethernet, network uses
thick (3/
8-inch) coaxial cable (with 50-ohm
impedance) for the network backbone. The
50-ohm cable is specially designed for this
version of Ethernet, but standard thick
coaxial cable can also be used.
A thick Ethernet network uses the following
components:
I Ethernet NICs to mediate between
node and network (one per node)
I Thick coaxial cable for trunk cable
segments (with nodes attached) or for
Thick Ethernet
Thick Ethernet Components
LAYERS INVOLVED IN HANDLING BLUE BOOK
AND 802.3 ETHERNET PACKETS


Prev Page 374 Next

360
Ethernet
link segments (between repeaters, and
with no nodes attached)
I Transceivers to attach to the trunk seg-
ment and to do the required conver-
sions when the node transmits or
receives (one per node)
I Transceiver, or drop, cable with DIX
connectors on each end, to connect the
NIC in the node to the transceiver
attached to the trunk segment (one
per node)
I N-series barrel connectors, to connect
pieces of cable in the trunk segments
(the fewer the better)
I N-series terminators, to terminate one
end of a trunk segment (one per trunk
segment)
I N-series grounded terminators, to ter-
minate and ground one end of a trunk
segment (one per trunk segment)
I Repeaters (optional), to extend the
network by regenerating the signal
before passing it on
The thick cable is relatively difficult to
manage and install. Most networks that use
thick cable use it as the network backbone,
which is not expected to change. The nodes
in the network are attached using additional
cable, called drop cable or transceiver cable.
The following configuration rules and
restrictions apply for thick Ethernet.
I The maximum length of a trunk
segment is 500 meters (1,640 feet).
Thick Ethernet Configuration
I The network trunk can have at most
five segments, for a total trunk of
2,500 meters (8,200 feet). Of these five
cable segments, up to two can be link
segments (without nodes attached) and
up to three can be trunk segments
(with nodes attached).
I
Within a thick coaxial trunk segment,
you can use N-series barrel connectors
to link shorter pieces of cable. You can
use repeaters to connect two segments
into a longer network trunk. A
repeater counts as a node on each of
the segments the repeater connects.
I You can have at most 100 nodes
(including repeaters) attached to each
trunk cable segment.
I A thick Ethernet network can have at
most 300 nodes, of which 8 will actu-
ally be repeaters.
I Each trunk segment must be termi-
nated at one end; the segment must
also be terminated and grounded at the
other end. When using thick coaxial
cable, this is accomplished using
N-series terminators, which are con-
nected to the male N-series connectors
at each end of the trunk segment.
I Nodes are connected to the trunk cable
using a transceiver cable from an AUI,
or DIX, connector on the NIC to an
AUI connector on a transceiver. The
male connector attaches to the NIC
and the female connector to the
transceiver.


Prev Page 375 Next

Ethernet
361
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
I The transceiver is connected to the
trunk cable with a vampire tap or with
an N-series T-connector.
I Transceivers must be at least 2.5
meters (8 feet) apart on the trunk,
although the machines themselves can
be closer together.
I The transceiver cable can be at most
50 meters (165 feet) long, which is the
maximum distance a node can be from
the network cable trunk.
The figure "Major components of a thick
Ethernet network" shows an example of a
thick Ethernet network.
A 10Base2, or thin Ethernet, network uses
thin (3/16-inch) coaxial cable (with 50-ohm)
impedance for the network backbone. Thin
coaxial cable is much easier to prepare and
install than thick Ethernet cable.
A thin Ethernet network uses the following
components:
I Ethernet NICs, containing a trans-
ceiver, to mediate between node and
network (one per node)
I Thin coaxial cable for trunk cable
segments
I BNC barrel connectors, to connect
pieces of cable in the trunk segments
(the fewer the better)
I BNC T-connectors, to attach a node
to the network (one per node)
I BNC terminators, to terminate one
end of a trunk segment (one per trunk
segment)
I BNC grounded terminators, to termi-
nate and ground one end of a trunk
segment (one per trunk segment)
I Repeaters (optional), to extend the
network by regenerating the signal
before passing it on
The following configuration rules and
restrictions apply for thin Ethernet:
I Each trunk segment can be at most
185 meters (607 feet). Each trunk seg-
ment can consist of multiple pieces
of cable, linked using BNC barrel
connectors.
I The network trunk can have at most
five segments, for a total trunk of 925
meters (3,035 feet). Of these five cable
segments, up to two can be link seg-
ments (those with no nodes attached)
and up to three can be trunk segments
(without nodes attached).
I You can use repeaters to connect two
segments into a longer network trunk.
A repeater counts as a node on each of
the segments the repeater connects.
I You can have at most 30 nodes
(including repeaters) attached to each
trunk cable segment.
I A thin Ethernet network can have at
most 90 nodes, of which 8 will actu-
ally be repeaters.
Thin Ethernet
Thin Ethernet Components
Thin Ethernet Configuration


Prev Page 376 Next

362
Ethernet
MAJOR COMPONENTS OF A THICK ETHERNET NETWORK


Prev Page 377 Next

Ethernet
363
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
I Each trunk segment must be termi-
nated at one end; the segment must
also be terminated and grounded at the
other end using BNC terminators,
which are connected to the male BNC
connectors at each end of the trunk
segment.
I
Nodes are connected to the trunk cable
using a BNC T-connector that is
attached to the NIC.
I T-connectors must be at least 0.5 meter
(1.6 feet) apart on the trunk, although
the machines themselves can be closer
together.
The figure "Major components of a thin
Ethernet network" shows an example of a
thin Ethernet network.
You can combine thin and thick coaxial
cable in the same Ethernet network, pro-
vided that the network elements meet the
appropriate cable specifications. This
approach can be less expensive than a
pure thick Ethernet configuration and
more robust than a pure thin Ethernet
configuration.
One approach is to combine thick and
thin coaxial cable within a trunk segment.
In this case, the connection is made using
hybrid (BNC/N-series) adapters. One end
of the adapter is a BNC connection and the
other end is an N-series connection. Two
versions of this adapter are available: one
has female connections at either end, and
the other has male connections.
Hybrid Ethernet
When thin and thick coaxial cables are
combined within the same segment, you
need a formula to determine the amount of
each type of cable you can use. The follow-
ing formula assumes that no trunk segment
is longer than 500 meters (1,640 feet):
(1,640 - Len)/3.28 = MaxThinCoax
where Len is the length of the trunk segment
and MaxThinCoax represents the maximum
length of thin coaxial cable you can use in
the segment.
You can also build a network trunk using
thin and thick trunk segments. In this case,
the transition is made at the repeaters. Each
segment must meet the specifications for
that type of cable, just as if the entire trunk
were made of the same type of cable.
As with thin or thick Ethernet segments,
each end of a hybrid segment must be termi-
nated. The terminator must match the type
of cable at the end. Thus, if one end of the
segment ends in thin coaxial and the other
ends in thick coaxial, you need a BNC ter-
minator at the first end and an N-series
terminator at the second end. You can
ground either of the ends.
Note that all the cable used in both thick
and thin Ethernet networks has the same
impedance: 50 ohms. This is one reason why
it is relatively easy to combine thin and thick
Ethernet segments.
Several companies have developed fast
Ethernets, which are implementations capa-
ble of 100 Mbps transmission speeds over
Trends: Fast Ethernet


Prev Page 378 Next

364
Ethernet
MAJOR COMPONENTS OF A THIN ETHERNET NETWORK


Prev Page 379 Next

Ethernet
365
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
UTP cable. These implementations are also
known as 100 Mbps Ethernets.
Two fast Ethernet variants were
accepted as official standards in June 1995.
100BaseVG, developed by Hewlett-Packard
and several other vendors, was recently
accepted as a standard by the IEEE 802.12
study group. On the same day, several vari-
ants of Grand Junction's 100BaseT were
accepted as extensions of the 802.3 10BaseT
standard. The variants are: 100BaseFX (for
fiber optic cable), 100BaseT4 (for connec-
tions with four available wire pairs), and
100BaseTX (for high-quality Category 5
cable).
In order to achieve such high speeds,
developers have found it necessary to take
liberties with certain Ethernet features, as
follows:
Access method: A major controversy con-
cerns the access method to be used.
Hardware vendors have formed camps
behind the HP and Grand Junction
versions, depending partly on whether
they want to retain the familiar
CSMA/CD. HP's 100BaseVG uses
demand priority as its media-access
method. This strategy involves packet
switching and takes place in the hubs
that serve to concentrate nodes on a
twisted-pair network. Grand Junc-
tion's 100BaseT uses CSMA/CD.
Cable Type: Current versions of twisted-
pair Ethernet run on cable that meets
the TIA-568 standards for Category 3
cable or higher. Category 3 cable is
rated for transmission speeds of up
to 10 Mbps, and standard Ethernet
requires two pairs of cable-one pair
for each direction. 100BaseT Ethernet
requires either four pairs of Category
3 cable or else two pairs of Category 5
cable (which is rated for 100 Mbps
speeds). 100BaseVG uses special sig-
naling methods, and so it can use ordi-
nary Category 3 cable.
NICs: Cards that support a 100 Mbps
Ethernet must be capable of switching
to the slower 10 Mbps speed, and
must be able to detect when it is neces-
sary to do so.
Fast Ethernet cards send a fast link pulse
(FLP) signal to indicate that they are capable
of 100 Mbps transmission. If this signal is
not detected, it is assumed that the node
is an ordinary (10 Mbps) one.
Other proposed features, such as the
frame format and configuration restrictions,
are the same as for the current 802.3
Ethernet.
An isochronous transmission is one that
occurs at a constant rate. This is required,
for example, when sending voice or video,
since the information could become unintel-
ligible if sent at varying speeds or with
pauses in mid-transmission. Such time-
dependent transmissions are not possible
with ordinary Ethernet-largely because the
media access method (MAC) is probabilistic
and is not designed for constant activity.
To make it possible to transmit voice and
video over Ethernet networks, National
Semiconductor has submitted specifications
for isoENET-an isochronous version of
Ethernet-to the IEEE 802.9 committee.
Isochronous Ethernet


Prev Page 380 Next

366
Ethernet Meltdown
802.9 is the committee that deals with the
integration of voice and data (IVD).
The isoENET specs support transmissions
using ISDN (Integrated Services Digital Net-
work) signaling methods-but running over
Category 3 UTP (unshielded twisted pair)
cable. IsoENET's 16 Mbps bandwidth is
broken into two major components. In addi-
tion to the 10 Mbps bandwidth for ordinary
Ethernet transmissions, isoENET supports
up to 96 B channels, each with a 64 kbps
capacity-for a total throughput of about
6 Mbps-for the isochronous part of the
transmission.
Ethernet networks offer the following
advantages:
I Good for networks in which traffic
is heavy only occasionally or in
which traffic consists of a few long
transmissions.
I Easy to install.
I Technology is well-known and
thoroughly tested.
I Moderate costs.
I Flexible cabling, especially when
using twisted-pair cable.
Ethernet networks have the following
disadvantages:
I Heavy traffic can slow down a net-
work that uses a contention access
system such as CSMA/CD. Such con-
gestion is less likely to be a problem
Advantages of Ethernet
Disadvantages of Ethernet
with the 100 Mbps Ethernets-at least
until the traffic catches up with the
greater bandwidth.
I
Since all nodes are connected to the
main cable in most Ethernet networks,
a break in this cable can bring down
the entire network.
I Troubleshooting is more difficult with
a bus topology.
I
Room for incompatibilities because of
frame structure (such as 802.3 versus
Blue Book Ethernet).
SEE ALSO
100BaseT; 100BaseVG; isoENET
BROADER CATEGOR Y
Network Architecture
COMPARE
ARCnet; ATM; FDDI; Token Ring
MEthernet Meltdown
A situation in which traffic on an Ethernet
network approaches or reaches saturation
(maximum capacity). This can happen, for
example, if a packet is echoed repeatedly.
M
EtherTalk
EtherTalk is the driver used to communicate
between the Macintosh and an Ethernet
network interface card. It is Apple's
Ethernet implementation for the Apple-
Talk environment.
Two versions of EtherTalk have been
developed:
I EtherTalk Phase 1 is based on the
Ethernet 2 version, also known as Blue
Book Ethernet.


Prev Page 381 Next

Event Reporting
367
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
I EtherTalk Phase 2 is based on the IEEE
802.3 Ethernet variant.
BROADER CATEGORIES
AppleTalk; Ethernet
COMPARE
ARCTalk; LocalTalk; TokenTalk
M
ETR (Early Token Release)
ETR is a frame, or packet, control process
used in 16 megabit per second (Mbps) token
ring networks. ETR makes it possible for
multiple packets to be moving in the ring at
once, even with just a single token for
packet control.
Ordinarily in a token ring network, only
the node with the token can send a packet,
so that only one packet is moving around
the network at any one time. This packet
travels around the ring. Each node passes
the packet on, and the destination node
reads the packet. When the packet returns
to the sender (with acknowledgment and
verification of its receipt), that node strips
the packet and passes the token to the
next active node on the ring.
With ETR, the sender releases the token
immediately after releasing its packet. The
next node on the ring sends the packet on.
Since this node now has the token, the node M
can send its own packet. Immediately after
sending the packet, the node releases the
token. Successive nodes pass on whatever
packets they receive, and they send their
own packets (if they have any to send) when
the token reaches them.
Note that ETR allows multiple packets
on the network, but that there is only one
token on the network at any time.
BROADER CATEGORIES
Token Passing; Token Ring
M
ETSI (European Telecommunications
Standards Institute)
A European standards committee that has
defined a subset of ISDN's proposed func-
tionality for use in Europe. This variant is
known as EuroISDN and is analogous to
the National ISDN versions (NI-1, NI-2,
and a planned NI-3) developed in the United
States. The ETSI is also looking into specify-
ing guidelines for providing interoperability
between EuroISDN and National ISDN.
M
European Academic and Research
Network (EARN)
A European network that provides file trans-
fer and e-mail (electronic mail) services for
universities and research institutions.
M
European Electronic Mail Association
(EEMA)
A European association of developers and
vendors of electronic mail products. The
EMA (Electronic Mail Association) is the
counterpart in the United States.
Event Reporting
In network management, a data-gathering
method in which agents report on the status
of the objects under the agents' purview. The
agent generates a report containing the rele-
vant information and sends this report to the
management package. This is in contrast to
polling, in which the management program


Prev Page 382 Next

368
eWorld
periodically requests such reports from
agents.
M
eWorld
An online service newly developed by Apple
for Macintosh users. eWorld is based on
AOL (America Online) software. It uses a
city as the metaphor for its graphics-based
interface. In this interface, city locations
(such as buildings or kiosks) provide access
to the available services. Currently, eWorld
supports Internet access only for e-mail, but
other services are planned.
SEE ALSO
AOL; CompuServe; Prodigy
FOR INFORMATION
Call 800-775-4556
M
EWOS (European Workshop for Open
Systems)
One of three regional workshops for imple-
menters of the OSI Reference Model. The
other two are AOW (Asia and Oceania
Workshop) and OIW (OSI Implementers
Workshop).
M
Exchange
In telephone communications, an exchange
is an area serviced by a central office, or CO.
An exchange consists of a sequential block
of phone numbers, each associated with
the same three-digit value (known as the
exchange ID, or XID).
Each exchange in North America is char-
acterized by an office class and a name. The
table "North America Exchange Classes and
Names" summarizes how the classes are
defined.
M
Exchange Carrier
A local exchange carrier (LEC), which is a
company that provides telecommunications
services within an exchange, or LATA (local
access and transport area).
MExpansion Bus
A set of slots, such as those on a mother-
board, into which expansion cards can be
plugged in order to provide the computer
with additional capabilities and access to
external devices.
M
Expansion Chassis
A structure that includes a backplane
(circuit board with slots for other boards)
and a power supply. The chassis may be
closed and self-standing, or it may be rack
mountable for installation into a larger
component.
M
Explorer Frame
In networks that use source routing, such
as IBM Token Ring networks, an explorer
frame is used to determine a route from the
source node to a destination. An explorer
frame is also known as a discovery packet,
particularly in the Internet community.
There are two types of explorer frames:
I An all-routes explorer frame explores
all possible routes between source and
destination


Prev Page 383 Next

Extensible MIB
369
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
I A spanning-tree explorer frame fol-
lows only routes on the spanning tree
for the network. (A spanning tree is an
optimal set of paths for all possible
connections in a network.)
M
Extended Addressing
In AppleTalk Phase 2, extended addressing
is a scheme that assigns an 8-bit node num-
ber and a 16-bit network number to each
station. Extended addressing allows for
up to 16 million (224) nodes on a single
network.
This is in contrast to the nonextended
addressing used in AppleTalk Phase 1 net-
works and also in networks that use a Local-
Talk architecture. Nonextended addressing
uses just the 8-bit node number, which limits
networks to 254 nodes (not 256, because
two of the node numbers are reserved).
Packets for extended networks use the
long DDP packet format; packets for non-
extended networks use the short DDP
packet format, which omits network address
bytes (since these are either undefined or 0).
M
Extensible MIB
In an SNMP environment, a MIB for which
a vendor can define new variables when
implementing the MIB.
SEE ALSO
MIB (Management Information Base);
SNMP (Simple Network Management
Protocol)
NOR TH AMERICA EXCHANGE CLASSES AND NAMES
EXCHANGE CLASS
NAME
1
2
3
4
4P
4X
5
5R
Regional centers (RCs) or points (RPs). These have the largest domains: a dozen
or so cover all of North America. The class 1 offices are all connected directly to
each other.
Sectional centers (SCs) or points (SPs).
Primary centers (PCs) or points (PPs).
Toll centers (TCs).
Toll points (TPs).
Intermediate points (IPs). These are used only with digital exchanges, and are
designed to connect to remote switching units (RSUs).
End offices. These are owned by local telephone companies. Ownership of the
broader centers varies. Individual subscribers are connected to class 5 offices,
of which there are many thousand in North America.
End offices with remote switching capabilities.


Prev Page 384 Next

370
Extensible MIB


Prev Page 385 Next



Prev Page 386 Next

372
Extensible MIB


Prev Page 387 Next

FF


Prev Page 388 Next

372
Facility
FM
Facility
In telephone communications, a transmis-
sion link between two locations, or stations.
In an X.25 packet, a facility is a field
through which users can request special
services from the network.
M
Facility Bypass
In telecommunications, a communication
strategy that bypasses the telephone com-
pany's central office. For example, wireless
transmissions might use facility bypass.
M
Facility Data Link (FDL)
SEE
FDL (Facility Data Link)
M
Fading
In electrical or wireless signaling, fading is
the decrease in the signal's strength because
of any of the following:
I Obstruction of the transmitter's or the
receiver's antenna
I Interference (from other signals or
from atmospheric conditions)
I Increased distance from the transmis-
sion source
Fading is sometimes referred to as just
fade, as in fade margin. The fade margin
refers to the amount of signal (in decibels)
that can be lost before the signal becomes
unintelligible.
M
FADU (File Access Data Unit)
In the OSI's FTAM (File Transfer, Access,
and Management) service, a packet that
contains information about accessing a
directory tree in the file system.
M
Fail-Safe System
A computer system that is designed to keep
operating, without losing data, when part of
the system seriously malfunctions or fails
completely.
M
Fail-Soft System
A computer system that is designed to fail
gracefully, with the minimum amount of
data or program destruction, when part of
the system malfunctions. Fail-soft systems
close down nonessential functions and oper-
ate at a reduced capacity until the problem
has been resolved.
MFake Root
In Novell's NetWare versions 3.x and 4.x, a
fake root is a drive mapping to a subdirec-
tory that makes the subdirectory appear to
be the root directory.
A fake root allows you to install pro-
grams into subdirectories, even though they
insist on executing in the root directory.
With the programs in a subdirectory, admin-
istrators can be more specific about where
they allow users to have rights, and avoid
granting rights at the true root of the
volume.
Fake roots are not allowed in all environ-
ments. For example, fake roots cannot be


Prev Page 389 Next

Fast Ethernet
373
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
used with OS/2 clients. When a fake root is
used, there are also restrictions on how cer-
tain commands work and on how certain
actions-for example, returning to the origi-
nal (non-fake) root-must be performed.
BROADER CATEGOR Y
NetWare
M
FAL (File Access Listener)
In Digital Equipment Company's DECnet
environment, a program that implements the
DAP (Data Access Protocol) and that can
accept remote requests from processes that
use DAP.
MFall Time
The amount of time it takes an electrical sig-
nal to go from 90 percent of its level down
to 10 percent. This value is important,
because it helps set an upper limit on the
maximum transmission speed that can be
supported. Compare it with rise time.
M
Fanout
In communications and signaling, a configu-
ration in which there are more output lines
than input lines.
MFAQ (Frequently Asked Questions)
In the Internet community, FAQ is a compi-
lation of the most commonly asked ques-
tions, with answers, about any of dozens of
topics. Many of these questions might be
asked by newcomers, who may know little
or nothing about a topic.
FAQs are posted in order to minimize the
number of users who actually do ask the
questions. Users can download and read the
answers at their leisure, rather than tying
up the lines by mailing these questions
across the Internet and waiting for the
answers to come pouring in.
FAQs can be found in archives on the
Internet, and will have names such as
disney-faq/disneyland, audio-faq/part1, or
usenet-faq/part1. In FAQ archives, you can
find a variety of information, such as where
to look for old, out-of-print Disney videos,
what to listen for when evaluating speakers
(the electronic kind), and so on.
M
Far End Block Error (FEBE)
In broadband ISDN (BISDN) networks, an
error reported to the sender by the receiver
when the receiver's computed checksum
result does not match the sender's checksum.
M
Far End Receive Failure (FERF)
In broadband ISDN (BISDN) networks, a
signal sent upstream to indicate that an error
has been detected downstream. An FERF
might be sent, for example, because a desti-
nation has reported an error.
M
Fast Ethernet
Any of several Ethernet variants based on an
approach developed by Grand Junction and
others. The official name for this brand of
Ethernet is 100BaseT (for twisted pair,
which refers to the type of cable), and there
are actually three variants, as described in
the article "100BaseT."


Prev Page 390 Next

374
Fastconnect Circuit Switching
The term is also used to refer to any 100
Mbps Ethernet implementation-for exam-
ple, 100BaseT or 100BaseVG. Finally,
fastEthernet was the name for a now-
defunct product.
M
Fastconnect Circuit Switching
The use of fast, electronic switching to
establish a path (circuit) between two
stations.
MFastPath
A high-speed gateway between AppleTalk
and Ethernet networks.
MFAT (File Allocation Table)
The FAT (file allocation table) is where DOS
keeps its information about all the files on a Fault Detection and Assessment
partition and about the disk location of all
the blocks that make up each file. Because
losing a FAT can be fatal in the PC world,
DOS maintains a second copy of the FAT.
Some network operating systems, such as
NetWare, also use FATs as part of their file
handling. For example, NetWare uses a
directory entry table (DET) and a FAT.
Access to the FAT is through the DET.
The total storage a FAT can map depends
on the size of each block (or allocation unit).
Hard disk blocks can be 4, 8, 16, 32, or 64
kilobytes (KB) each.
The number of blocks that can be covered
by the FAT is constant, at least for all but
the earliest versions of the FAT. Large blocks
are good for large files; smaller blocks are
best for lots of small files.
Various tricks can be used to speed up
access to the FAT, including caching and
indexing the FAT. Caching the FAT involves
storing it in chip memory (RAM) for faster
access. Indexing information in a FAT
can be accomplished by using a hashing
function.
M
Fault
A break or other abnormal condition in
a communications link. A fault generally
requires immediate attention. The fault may
be physical or logical.
M
Fault Management
One of five basic OSI network management
tasks specified by the ISO and CCITT, fault
management is used to detect, diagnose, and
correct faults on a network.
A network management package can detect
faults by having nodes report when a fault
occurs, as well as by polling all nodes peri-
odically. Both capabilities are necessary for
thorough fault management. It may not be
possible to get reliable reports about certain
types of faults, such as one that causes an
entire network to go down. For such cases,
polling will provide at least the negative
information of no response to a poll.
On the other hand, polling uses band-
width that could be used for transmitting
information. As in the real world, the more
time spent on administrative work (polling),
the less opportunity for doing real work
(transmitting information). The value of the
information obtained through polling must
be weighed against the loss of bandwidth.


Prev Page 391 Next

Fault Point
375
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
The bandwidth consumed by polling
depends also on the complexity of the poll-
ing method. For example, a simple method
sends a signal and waits for an echo to
acknowledge that the channel is open. All
network management environments include
facilities for echo polling. More complex
polling may check for more details, such
as whether the node has something to send
and whether a higher-priority level is
requested.
When a fault is detected, the network
management package must assess the fault
to determine whether it is necessary to track
it down and correct it immediately. Certain
types of faults affect or shut down vital net-
work services, and these faults must be dealt
with as soon as possible. Other faults may
involve only a path between locations, and
they may not be crucial because alternate
paths exist.
To determine the type of fault and its
locations, the network management package
may need to do some testing. For example, if
a poll does not get the expected echo, the
management package needs to determine
whether the fault is in the poller, the pollee,
or the link between them. This may require
signal monitoring or loopback testing.
Once the fault has been detected, identified,
and located, measures must be taken to cor-
rect it. In some cases, such as when there is
redundancy in the system, the management
package may be able to correct the fault
automatically. More likely, the network
administrator or engineer will need to inter-
vene in order to correct the fault. The ease
with which this happens depends on the
reliability of the detection and diagnosis,
and on the type of information provided
about the fault.
The fault-management system must be
able to trace faults through the network and
carry out diagnostic tests. Fault correction
requires help from the configuration man-
agement domain.
To collect the information necessary to
detect and report faults, fault-management
systems use either of two families of proto-
cols: the older SNMP (Simple Network
Management Protocol) or the OSI standard
CMIP (Common Management Information
Protocol).
Faults can be reported in various ways.
The simplest (and least informative) is an
auditory alarm signal, which merely alerts
the system administrator.
Actual information about the fault can
be reported as text, or through a graphical
interface that shows the network layout
schematically, with the fault located in this
diagram.
BROADER CATEGOR Y
Network Management
SEE ALSO
Accounting Management; Configuration
Management; Performance Manage-
ment; Security Management
MFault Point
In networking, a location at which some-
thing can go wrong. Fault points often tend
to be at connection locations.
Fault Correction
Fault Reporting


Prev Page 392 Next

376
Fault Tolerance
M
Fault Tolerance
Fault tolerance is a strategy for ensuring
continued operation of a network even
when certain kinds of faults arise. Fault-
tolerant networks require some sort of
redundant storage medium, power supply,
or system.
For example, a fault-tolerant cabling sys-
tem will include extra cables, in case one
cable is cut or otherwise damaged. A fault-
tolerant disk subsystem will include multiple
copies of data on separate disks and use sep-
arate channels to write each version.
In some configurations, it is possible to
remove and replace the malfunctioning com-
ponent (for example, a hard disk) without
shutting down the system. See the SFT (Sys-
tem Fault Tolerance) article for information
about Novell NetWare's fault-tolerant
features.
BROADER CATEGORIES
Data Protection; Security
M
Fax
A fax is a long-distance photocopy; it is a
reproduction of a text or graphics document
at a remote location. The document is
scanned (or already available in digitized
form), encoded into a standard format for
faxes, transmitted over telephone or private
lines, and printed (or stored) at the receiving
end. Telecopy and telefax are other terms for
fax. The figure "The fax transmission pro-
cess" illustates how a fax is sent.
Fax images have resolutions that range
from about 100 200 (vertical horizontal)
dots per inch (dpi) to about 400 400 dpi.
The CCITT has formulated fax format
and transmission standards, referred to as
Groups 14, which represent a range of sig-
naling methods and formats, as follows:
I Group 1 uses frequency modulation of
analog signals and supports only slow
transmission speeds (6 minutes per
page). Group 1 offers low (100 dpi)
resolution.
I Group 2 uses both frequency and
amplitude modulation to achieve
higher speeds (between 2 and 3 min-
utes per page). Group 2 also offers low
(100 dpi) resolution.
I Group 3 uses quadrature amplitude
modulation (QAM) and data compres-
sion to increase transmission speeds to
about one page per minute. Group 3
supports various automatic features
and offers 200 dpi resolution. Com-
mercially available fax machines sup-
port at least the Group 3 format.
I Group 4 supports higher-speed digital
transmissions, so that a page can be
transmitted in about 20 seconds.
Group 4 offers 200 or 400 dpi resolu-
tion. Three classes are distinguished
under the Group 4 format (which is
not yet in wide use).
SEE ALSO
Modulation


Prev Page 393 Next

Fax Device
377
a
b
c
d
e
f
g
h
i
jk
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
M
Fax Device
A fax device can be used to send and receive
faxes on a network, under the control of
a fax server. This may be a machine or a
board. Machines may use thermal or plain
paper.
In general, thermal paper comes in rolls,
fades and cracks quickly, and must be cut as
the fax leaves the machine. The main (and
THE FAX TRANSMISSION PROCESS
An electronic or a physical document is
created. An electronic document can be
entered as text at the keyboard or
scanned into the computer. Aphysical
document may be created by hand or by
printing an electronic document.
An electronic or a physical document is
created. An electronic document can be
entered as text at the keyboard or
scanned into the computer. Aphysical
document may be created by hand or by
printing an electronic document. Aphysical
document may be created by hand or by
printing an electronic document.
An electronic or a physical document is
created. An electronic document can be
entered as text at the keyboard or
scanned into the computer. Aphysical
document may be created by hand or by
printing an electronic document.
An electronic or a physical document is
created. An electronic document can be
entered as text at the keyboard or
scanned into the computer. Aphysical
document may be created by hand or by
printing an electronic document.
An electronic or a physical document is
created. An electronic document can be
entered as text at the keyboard or
scanned into the computer. Aphysical
document may be created by hand or by
printing an electronic document.
An electronic or a physical document is
created. An electronic document can be
entered as text at the keyboard or
scanned into the computer. Aphysical
document may be created by hand or by
printing an electronic document. Aphysical
document may be created by hand or by
printing an electronic document.
An electronic or a physical document is
created. An electronic document can be
entered as text at the keyboard or
scanned into the computer. Aphysical
document may be created by hand or by
printing an electronic document.
An electronic or a physical document is
created. An electronic document can be
entered as text at the keyboard or
scanned into the computer. Aphysical
document may be created by hand or by
printing an electronic document.
An electronic or a physical document is
created. An electronic document can be
entered as text at the keyboard or
scanned into the computer. A physical
document may be created by hand or by
printing an electronic document.
An electronic or a physical document is
created. An electronic document can be
entered as text at the keyboard or
scanned into the computer. Aphysical
document may be created by hand or by
printing an electronic document.
An electronic or a physical document is
created. An electronic document can be
entered as text at the keyboard or
scanned into the computer. Aphysical
document may be created by hand or by
printing an electronic document. Aphysical
document may be created by hand or by
printing an electronic document.
An electronic or a physical document is
created. An electronic document can be
entered as text at the keyboard or
scanned into the computer. Aphysical
document may be created by hand or by
printing an electronic document.
An electronic or a physical document is
created. An electronic document can be
entered as text at the keyboard or
scanned into the computer. Aphysical
document may be created by hand or by
printing an electronic document.
The fax format is transmitted over
telephone lines, using either a fax board
or a fax machine. At the receiving end, a
fax board or machine converts the fax
format back into a document or image.
This may be printed immediately (as on
a fax machine) or stored internally as a
file (as with fax boards).
The document is converted to a fax
format. For electronic documents, the
conversion is done by a fax board; for
physical documents, this is done by
scanning them into a fax machine and
performing the conversion during the
scanning process.
An electronic or a physical document is
created. An electronic document can be
entered as text at the keyboard or
scanned into the computer. Aphysical
document may be created by hand or by
printing an electronic document.
An electronic or a physical document is
created. An electronic document can be
entered as text at the keyboard or
scanned into the computer. Aphysical
document may be created by hand or by


entered as text at the keyboard or
scanned into the computer. Aphysical
document may be created by hand or by
printing an electronic document.
An electronic or a physical document is
created. An electronic document can be
entered as text at the keyboard or
scanned into the computer. Aphysical
document may be created by hand or by
printing an electronic document.
An electronic or a physical document is
An electronic or a physical document is
created. An electronic document can be
entered as text at the keyboard or
scanned into the computer. Aphysical
document may be created by hand or by
printing an electronic document.
An electronic or a physical document is
created. An electronic document can be
entered as text at the keyboard or
scanned into the computer. Aphysical
document may be created by hand or by


entered as text at the keyboard or
scanned into the computer. Aphysical
document may be created by hand or by
printing an electronic document.
An electronic or a physical document is
created. An electronic document can be
entered as text at the keyboard or
scanned into the computer. Aphysical
document may be created by hand or by
printing an electronic document.
An electronic or a physical document is


Prev Page 394 Next

378
FBE (Free Buffer Enquiry)
only) advantage of thermal paper fax
machines is price.
Fax boards can generally accept text or
graphics files (in the appropriate format),
can convert these into fax format, and can
transmit the resulting information. Fax
boards can also receive faxes and convert
them to the appropriate form for use.
Because fax boards have no paper supply of
their own, most boards can send their files
to a printer for hard copy.
Although the speed and resolution capa-
bilities for most fax machines are similar-
thanks, in part, to the CCITT fax stan-
dards-there are certain considerations
when selecting a fax device for use on a
network.
For example, if your network receives
many faxes daily, you will not want to use
a thermal paper fax machine that insists on
printing every fax received. On a busy day,
there might be a 100-foot roll of faxes to
wade through (literally) in order to find your
fax. For a network, you will probably want
the fax device to suppress printing (if
requested) and pass an electronic version
of the received fax to the appropriate
program.
BROADER CATEGOR Y
Peripheral
SEE ALSO
Modem; Server; Fax
M
FBE (Free Buffer Enquiry)
A field in an ARCnet frame.
SEE
ARCnet
MFC (Frame Control)
A field in a token ring data packet, or frame.
The FC value tells whether the frame is a
MAC-layer management packet or whether
it is carrying LLC (logical-link control) data.
M
FCC (Federal Communications
Commission)
A federal regulatory agency that develops
and publishes guidelines to govern the oper-
ation of communications and other electrical
equipment in the United States.
Perhaps the best-known FCC regulations
are those that define and govern class A and
class B devices, and those that allocate the
electromagnetic spectrum. The device certifi-
cations are based on the amount of radio
frequency interference (RFI) the device may
cause for other devices in the vicinity.
Class A certification is less stringent, and
it is assigned to equipment for use in busi-
ness contexts. The more stringent class B
certification applies to devices that are used
in the home.
The FCC also allocates portions of the
electromagnetic spectrum for particular
uses, such as the following:
I The frequency band between 88 and
108 megahertz (MHz) is allocated for
FM radio broadcasting.
I The bands between 54 and 88 MHz
and between 174 and 216 MHz are
allocated for VHF television.


Prev Page 395 Next

FDDI (Fiber Distributed Data Interface)
379
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
r
s
t
u
v
w
x
y
z
I The band between 470 and 638 MHz
is allocated for UHF television.
I Bands in the 4, 6, and 11 gigahertz
(AHz) ranges have been allocated for
long-haul telecommunications using a
common carrier.
I Bands in the 18 and 23 GHz ranges
have been allocated for short-haul
transmissions, such as those in private
networks.
M
FCS (Fiber Channel Standard)
The specifications for optical fiber in the
FDDI (Fiber Distributed Data Interface)
network architecture.
MFCS (Frame Check Sequence)
In network or other transmissions, a value
that is used to check for errors in a transmit-
ted message. The FCS value is determined
before sending the message, and it is stored
in the packet's FCS field. If the new FCS
value computed from the received packet
does not match the original, a transmission
error has occurred.
SEE ALSO
Error Detection and Correction
M
FDDI (Fiber Distributed Data
Interface)
FDDI is a proposed ANSI standard specifi-
cation (X3T9.5) for a network architecture
that is designed to use fiber-optic lines at
very high speeds.
An FDDI network has the following
characteristics:
I Uses multimode or single-mode fiber-
optic cable.
I
Supports transmission speeds of up to
100 megabits per second (Mbps).
I
Uses a ring topology. Actually, FDDI
uses dual rings on which information
can travel in opposite directions.
I Uses token-passing as the media-access
method. However, in order to support
a high transmission rate, FDDI can
have multiple frames circulating the
ring at a time, just as with ETR (early
token release) in an ordinary Token
Ring network.
I Uses light, rather than electricity, to
encode signals.
I Uses a 4B/5B signal-encoding scheme.
This scheme transmits 5 bits for every
4 bits of information. (This means that
an FDDI network needs a clock speed
of 125 Mbps to support a 100 Mbps
transmission rate.) The actual bits are
encoded using an NRZ-I strategy.
I Uses an LED (light-emitting diode) or
a laser operating at a wavelength of
roughly 1,300 nanometers (nm). This
wavelength was chosen because it pro-
vides suitable performance even with
LEDs.
I
Supports up to 1,000 nodes on the
network.
I Supports a network span of up to 100
kilometers (62 miles).


Prev Page 396 Next

380
FDDI (Fiber Distributed Data Interface)
I Supports nodes up to 2 kilometers
(1.25 miles) apart when using
multimode cable and up to 40 kilome-
ters (25 miles) when using single-mode
cable.
I Supports a power budget (allowable
power loss) of 11 decibels (dB)
between nodes. This value means that
about 92 percent of the signal's power
can be lost between two nodes. (The
signal is at least partially regenerated
by the transceiver at each node.)
I Can handle packets from either the
LLC (logical-link control) sublayer
of the data-link layer or from the
network layer.
I Supports hybrid networks, which can
be created by attaching a subnetwork
(for example, a collection of stations
arranged in a star or a tree) to the ring
through a concentrator.
The figure "Context and properties of
FDDI" summarizes this architecture.
The FDDI architecture can be used for three
types of networks:
I In a backbone network, in which the
FDDI architecture connects multiple
networks. Optical fiber's very high
bandwidth makes FDDI ideal for such
applications.
I As a back-end network to connect
mainframes, minicomputers, and
peripherals. Again, the high bandwidth
makes FDDI attractive.
FDDI Applications
I As a front-end network to connect
special-purpose workstations (such
as graphics or engineering machines)
for very high-speed data transfer.
The FDDI standard consists of four docu-
ments: PMD, PHY, MAC, and SMT. Each
of which describes a different facet of the
architecture.
PMD represents the lowest sublayer sup-
ported by FDDI. This document specifies the
requirements for the optical power sources,
photodetectors, transceivers, MIC (medium
interface connector), and cabling. This is the
only optic (as opposed to electrical) level
and corresponds roughly to the lower parts
of the physical layer in the OSI Reference
Model.
The power source must be able to send
a signal of at least 25 microwatts (25 mil-
lionths of a watt) into the fiber. The photo-
detector, or light receptor, must be able to
pick up a signal as weak as 2 microwatts.
The MIC for FDDI connections serves as
the interface between the electrical and opti-
cal components of the architecture. This
connector was specially designed by ANSI
for FDDI and is also known as the FDDI
connector.
The cabling specified at this sublayer calls
for two rings running in opposite directions.
The primary ring is the main transmission
medium. A secondary ring provides redun-
dancy by making it possible to transmit the
data in the opposite direction if necessary.
When the primary ring is working properly,
the secondary ring is generally idle.
FDDI Documents
PMD (Physical Medium Dependent)


Prev Page 397 Next

FDDI (Fiber Distributed Data Interface)
381
a
b
c
d
e
f
g
h
i
j
k
l
m
n
o
p
q
rst
u
v
w
x
y
z
The PHY layer mediates between the MAC
layer above and the PMD layer below it.
Unlike the PMD layer, this is an electronic
layer. Signal-encoding and signal-decoding
schemes are defined at the PHY layer.
Functionally, this corresponds to the
PHY (Physical)
upper parts of the OSI Reference Model
physical layer.
The MAC layer defines the frame formats
and also the media-access method used by
the network. This corresponds to the lower
MAC (Media Access Control)
CONTEXT AND PROPER TIES OF FDDI
Context
Network Architectures
Electrical

Ethernet, ARCnet, etc.

Coaxial
Optical

FDDI
FDDI Properties
Medium
Multi-mode or single-mode optical fiber
Light source
LED or laser operating at approximately 1300 nm wavelength
Encoding scheme
4B/5B + NRZI
Topology
Dual rings, traveling in opposite directions

Access method
Token passing, but with multiple frames allowed
Data frame size
Maximum of 4500 data bytes plus 8+ bytes for a preamble
Layers
PMD optical, PHY, MAC, SMT
Performance
Supports transmission speeds of up to 100 Mbps

Can provide and maintain a guaranteed bandwidth

Supports up to 1000 nodes on the network

Supports a network span of up to 100 km

Supports nodes up to 2 km apart with multimode cable; up to 40 km with single-mode cable
Variants
FDDI-I and FDDI-II


Prev Page 398 Next

382
FDDI (Fiber Distributed Data Interface)
part of the OSI Reference Model data-link
layer. The MAC and PHY layers are imple-
mented directly in the FDDI chip set.
The MAC layer gets its data from the
LLC sublayer above it.
The SMT component monitors and manages
the node's activity. The SMT facility also
allocates the architecture's bandwidth as
required.
There are three elements to the SMT
component:
I Frame services generate frames for
diagnostics.
I Connection management (CMT)
controls access to the network.
I Ring management (RMT) trouble-
shoots the network.
If there is a fault in the primary ring, the
SMT facility redirects transmissions to use
the secondary ring around the faulty section.
This component can also use the secondary
ring to transmit data under certain condi-
tions, achieving a potential transmission rate
of 200 Mbps. This component has no coun-
terpart in the OSI Reference Model. SMT
capabilities may be implemented in hard-
ware or software.
The original FDDI specification (retroac-
tively named FDDI-I) called only for asyn-
chronous communications using packet-
switching. (Actually, there was a synchro-
nous traffic class in FDDI-I, but this did not
SMT (Station Management)
FDDI Versions
guarantee a uniform data stream as would
be required, for example, for voice or certain
video data.)
To handle voice, video, and multimedia
applications in real-time, a uniform data-
transmission capability was added in a revi-
sion that is generally known as FDDI-II, but
that is officially named hybrid ring control
(HRC) FDDI. This new capability uses
circuit-switching, so that FDDI-II supports
both packet- and circuit-switched services.
The figure "FDDI-I and FDDI-II organiza-
tion" shows the major differences between
the two versions.
The major structural additions to FDDI-II
are a medium access control element capable
of dealing with circuit-switched data and a
multiplexer capable of passing either packet-
or circuit-switched (that is, data, voice,
video, and so on) material to the physical
layer. This hybrid multiplexer (HMUX) gets
frames from both the MAC connected to the
LLC sublayer and from the isochronous
MAC, or IMAC, added in FDDI-II.
The IMAC interacts with one or more
circuit-switched multiplexers (CS-MUXs),
which are capable of delivering voice, video,
or any other kind of data that requires a