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

The Encyclopedia
of Networking

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 Düsseldorf I Soest

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

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.

[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

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.

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

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

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

&&
Symbols &
Numbers

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 (ö).
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, [email protected]
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).

4B/5B Encoding
3
M
µ(Mu)
Used as an abbreviation for the prefix micro,
as in µsec for microsecond and µm 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

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

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

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.

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

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

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.

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.

AA

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)

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

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

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.

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

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)

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

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.

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

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

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

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

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)

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

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

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).

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)

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)

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.)

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

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).

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

34
AppleTalk
THE APPLETALK PROTOCOL HIERARCHY
Please register!

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

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

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

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.

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

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.

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.

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.

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.

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

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

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

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

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

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

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.

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

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.

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

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

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).

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

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

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
[
\
]
^
_
à
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

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

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
}


Ç
ü
é
â
ä
à
å
ç
ê
ë
è
ï
î
ì
Ä
Å
É
æ
Æ
ô
ö
ò

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
û
ù
ÿ
Ö
Ü
¢
£
¥

á
í
ó
ú
ñ
Ñ
¡
«
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

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

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.

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

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

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

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

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

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

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.

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

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