Protocol Brief TCP/IP

TCP/IP Background
In the mid 1970s, the Defense Advanced Research Project Agency
(DARPA) was interested in providing packet-switched network commu-
nications between the many research institutions in the United States.
DARPA and other government organizations understood the potential
of packet-switched technology and were just beginning to discover that
virtually all companies with networks needed to support communica-
tion among dissimilar computer systems.

With the goal of heterogeneous connectivity in mind, DARPA funded

research by Stanford University and Bolt, Beranek, and Newman to
create a series of communication protocols. The result of that devel-
opment effort, completed in the late 1970s, was the Internet protocol

suite, of which the Transmission Control Protocol (TCP) and the

Internet Protocol (IP) are the two best-known members.

Use of the term “Internet” to refer to these protocols is appropriate,

because the Internet protocols were developed to operate across and
provide connectivity between already existing networks (for example,
the telephone network, dedicated links, and satellite circuits). The
design of the Internet protocols explicitly accounted for the fact that the
networks being tied together were heterogeneous in nature. They each
supported different speeds, error characteristics, data unit sizes, and
information formats.
TCP/IP

In the two decades since their invention, the heterogeneity of TCP/IP Technology
networks has expanded further with the deployment of Ethernet, This section describes technical aspects of TCP, IP, related pro-
Token Ring, Fiber Distributed Data Interface (FDDI), X.25, tocols, and the environments in which these protocols operate.
Frame Relay, Switched Multimegabit Data Service (SMDS), Because the primary focus of this document is routing (a layer
Integrated Services Digital Network (ISDN), and most recently, 3 function), the discussion of TCP (a layer 4 protocol) will be
Asynchronous Transfer Mode (ATM). The Internet protocols are relatively brief.
the best proven approach to internetworking this diverse range
of LAN and WAN technologies. TCP
TCP is a connection-oriented transport protocol that sends data
The Internet protocol suite includes not only lower-level specifi- as an unstructured stream of bytes. By using sequence numbers
cations (such as TCP and IP), but specifications for such com- and acknowledgment messages, TCP can provide a sending node
mon applications as electronic mail, terminal emulation, and file with delivery information about packets transmitted to a desti-
transfer. Figure 1 shows some of the more important Internet nation node. Where data has been lost in transit from source
protocols and their relationship to the OSI Reference Model. to destination, TCP can retransmit the data until either a time-
out condition is reached or until successful delivery has been
The Internet protocols are the most widely implemented multi- achieved. TCP can also recognize duplicate messages and will
vendor protocol suite in use today. Support for at least part discard them appropriately. If the sending computer is transmit-
of the Internet protocol suite is available from virtually every ting too fast for the receiving computer, TCP can employ flow
computer vendor. control mechanisms to slow data transfer. TCP can also commu-
nicate delivery information to the upper-layer protocols and
applications it supports.

Relationship of the Internet Protocol Suite to the OSI Reference Model

Figure 1

OSI
Reference
Model

7 Application NFS DNS, SNMP

FTP,
6 Presentation Telnet, XDR
SMTP
NetBIOS
5 Session RPC

4 Transport TCP, UDP

3 IP OSPF IGRP RIP INT, IS-IS E-IGRP

Network

ARP, RARP, SNAP

2 Data Link
LLC

1 Physical Many Physical Implementations

IP IP addressing supports three different network classes. Class A
IP is the primary layer 3 protocol in the Internet suite. In addi- networks are intended mainly for use with a few very large net-
tion to internetwork routing, IP provides error reporting and works, because they provide only 8 bits for the network address
fragmentation and reassembly of information units called field. Class B networks allocate 16 bits, and Class C networks
datagrams for transmission over networks with different max- allocate 24 bits for the network address field. Class C networks
imum data unit sizes. IP represents the heart of the Internet only provide 8 bits for the host field, however, so the number of
protocol suite. hosts per network may be a limiting factor. In all three cases, the
leftmost bit(s) indicate the network class. IP addresses are writ-
IP addresses are globally unique, 32-bit numbers assigned ten in dotted decimal format; for example, 34.0.0.1. Figure 2
by the Network Information Center. Globally unique addresses shows the address formats for Class A, B, and C IP networks.
permit IP networks anywhere in the world to communicate
with each other.

An IP address is divided into three parts. The first part des-

ignates the network address, the second part designates the
subnet address, and the third part designates the host address.

Class A, B, and C Address Formats

Figure 2

Class A 0

Network Host

Class B 1 0

Network Host

Class C 1 1 0

Network Host
TCP/IP

IP networks also can be divided into smaller units called subnet-

works or “subnets.” Subnets provide extra flexibility for the net-
work administrator. For example, assume that a network has
been assigned a Class A address and all the nodes on the network
use a Class A address. Further assume that the dotted decimal
representation of this network’s address is 34.0.0.0. (All zeros
in the host field of an address specify the entire network.) The
administrator can subdivide the network using subnetting. This
is done by “borrowing” bits from the host portion of the address
and using them as a subnet field, as depicted in Figure 3.

If the network administrator has chosen to use 8 bits of subnetting,

the second octet of a Class A IP address provides the subnet number.
In our example, address 34.1.0.0 refers to network 34, subnet 1;
address 34.2.0.0 refers to network 34, subnet 2, and so on.

Subnet Addresses
Figure 3

Class B Address: Before Subnetting

Class B 1 0

Network Host

Class B Address: After Subnetting

Class B 1 0

Network Subnet Host

The number of bits that can be borrowed for the subnet address As IP subnets have grown, administrators have looked for ways
varies. To specify how many bits are used and where they are to use their address space more efficiently. One of the tech-
located in the host field, IP provides subnet masks. Subnet niques that has resulted is called Variable Length Subnet Masks
masks use the same format and representation technique as (VLSM). With VLSM, a network administrator can use a long
IP addresses. Subnet masks have ones in all bits except those mask on networks with few hosts and a short mask on subnets
that specify the host field. For example, the subnet mask that with many hosts. However, this technique is more complex
specifies 8 bits of subnetting for Class A address 34.0.0.0 is than making them all one size, and addresses must be
255.255.0.0. The subnet mask that specifies 16 bits of subnetting assigned carefully.
for Class A address 34.0.0.0 is 255.255.255.0. Both of these sub-
net masks are pictured in Figure 4. Subnet masks can be passed Of course in order to use VLSM, a network administrator must
through a network on demand so that new nodes can learn how use a routing protocol that supports it. Cisco routers support
many bits of subnetting are being used on their network. VLSM with Open Shortest Path First (OSPF), Integrated Inter-
mediate System to Intermediate System (Integrated IS-IS),
Traditionally, all subnets of the same network number used the Enhanced Interior Gateway Routing Protocol (Enhanced IGRP),
same subnet mask. In other words, a network manager would and static routing.
choose an eight-bit mask for all subnets in the network. This
strategy is easy to manage for both network administrators and
routing protocols. However, this practice wastes address space
in some networks. Some subnets have many hosts and some
have only a few, but each consumes an entire subnet number.
Serial lines are the most extreme example, because each has
only two hosts that can be connected via a serial line subnet.

A Sample Subnet Mask

Figure 4

Class A
Address
0 0 1 0 0 0 1 0 ➞ 0 ➞ ➞ 0 ➞ ➞ 0 ➞ 34.0.0.0

Subnet
Mask:
8 Subnet ➞ 1 ➞ ➞ 1 ➞ ➞ 0 ➞ ➞ 0 ➞ 255.255.0.0
Bits

Class A
0 0 1 0 0 0 1 0 ➞ 0 ➞ ➞ 0 ➞ ➞ 0 ➞ 34.0.0.0
Address

Subnet
Mask:
➞ 1 ➞ ➞ 1 ➞ ➞ 1 ➞ ➞ 0 ➞ 255.255.255.0
16 Subnet
Bits
TCP/IP

On some media, such as IEEE 802 LANs, IP addresses are dynam- Routing in IP Environments
ically discovered through the use of two other members of the An “internet” is a group of interconnected networks. The
Internet protocol suite: Address Resolution Protocol (ARP) and Internet, on the other hand, is the collection of networks that
Reverse Address Resolution Protocol (RARP). ARP uses broadcast permits communication between most research institutions,
messages to determine the hardware (MAC layer) address corre- universities, and many other organizations around the world.
sponding to a particular network-layer address. ARP is suffi- Routers within the Internet are organized hierarchically. Some
ciently generic to allow use of IP with virtually any type of routers are used to move information through one particular
underlying media access mechanism. RARP uses broadcast mes- group of networks under the same administrative authority
sages to determine the network-layer address associated with a and control. (Such an entity is called an autonomous system.)
particular hardware address. RARP is especially important to Routers used for information exchange within autonomous sys-
diskless nodes, for which network-layer addresses usually are tems are called interior routers, and they use a variety of interior
unknown at boot time. gateway protocols (IGPs) to accomplish this end. Routers that
move information between autonomous systems are called exte-
rior routers; they use the Exterior Gateway Protocol (EGP) or
Border Gateway Protocol (BGP). The Internet architecture is
shown in Figure 5.

Representation of the Internet Architecture

Figure 5

Autonomous System Autonomous System

E

Enhanced IGRP EGP

RIP BGP
I OSPF
Integrated IS-IS

Macintosh

Token Token
Ring Ring
IBM PC

Ethernet Ethernet

I = Interior Routing Protocol

E = Exterior Routing Protocol
Routing protocols used with IP are dynamic in nature. Dynamic As we have seen, IP routing specifies that IP datagrams travel
routing requires the software in the routing devices to calculate through an internetwork one router hop at a time. The entire
routes. Dynamic routing algorithms adapt to changes in the net- route is not known at the outset of the journey. Instead, at each
work and automatically select the best routes. In contrast with stop, the next router hop is determined by matching the destina-
dynamic routing, static routing calls for routes to be established tion address within the datagram with an entry in the current
by the network administrator. Static routes do not change until node’s routing table. Each node’s involvement in the routing
the network administrator changes them. process consists only of forwarding packets based on internal
information. IP does not provide for error reporting back to the
IP routing tables consist of destination address/next hop pairs. source when routing anomalies occur. This task is left to another
In the sample routing table shown in Figure 6, the first entry is Internet protocol: the Internet Control Message Protocol (ICMP).
interpreted as meaning “to get to network 34.1.0.0 (subnet 1 on
network 34), the next stop is the node at address 54.34.23.12.” ICMP performs a number of tasks within an IP internetwork. In
addition to the principal reason for which it was created (report-
ing routing failures back to the source), ICMP provides a method
for testing node reachability across an internet (the ICMP Echo
and Reply messages), a method for increasing routing efficiency
(the ICMP Redirect message), a method for informing sources
that a datagram has exceeded its allocated time to exist within
an internet (the ICMP Time Exceeded message), and other help-
ful messages. All in all, ICMP is an integral part of any IP imple-
mentation, particularly those that run in routers.

An IP Routing Entry
Figure 6

Destination Next
Address Hop

34.1.0.0 54.34.23.12
78.2.0.0 54.34.23.12
147.9.5.0 .
17.12.0.0 .
. 54.32.12.10
. 54.32.12.10
. .
..
TCP/IP

Interior Routing Protocols Cisco has recently enhanced IGRP to handle the increasingly
Interior Routing Protocols or IGPs operate within autonomous large, mission-critical networks being designed today. This new
systems. The following sections provide brief descriptions of version of IGRP is called Enhanced IGRP. Enhanced IGRP com-
several IGPs that are currently popular in TCP/IP networks. bines the ease of use of traditional distance vector routing pro-
tocols with the fast rerouting capabilities of the newer link state
RIP routing protocols.
A discussion of routing protocols within an IP environment
must begin with the Routing Information Protocol (RIP). RIP Enhanced IGRP consumes significantly less bandwidth than
was developed by Xerox Corporation in the early 1980s for use IGRP because it is able to limit the exchange of routing infor-
in Xerox Network Systems (XNS) networks. Today, many PC mation to include only the changed information. In addition,
networks use routing protocols based on RIP. Enhanced IGRP is capable of handling AppleTalk and Novell
IPX routing information, as well as IP routing information.
RIP works well in small environments but has serious limita-
tions when used in larger internetworks. For example, RIP limits OSPF
the number of router hops between any two hosts in an internet OSPF was developed by the Internet Engineering Task Force
to 16. RIP is also slow to converge, meaning that it takes a rela- (IETF) as a replacement for RIP. OSPF is based on work started
tively long time for network changes to become known to all rout- by John McQuillan in the late 1970s and continued by Radia
ers. Finally, RIP determines the best path through an internet by Perlman and Digital Equipment Corporation (DEC) in the
looking only at the number of hops between the two end nodes. mid-1980s. Every major IP routing vendor supports OSPF.
This technique ignores differences in line speed, line utiliza-
OSPF is an intradomain, link state, hierarchical routing protocol.
tion, and all other metrics, many of which can be important
OSPF supports hierarchical routing within an autonomous sys-
factors in choosing the best path between two nodes. For this
tem. Autonomous systems can be divided into routing areas. A
reason, many companies with large internets are migrating
routing area is typically a collection of one or more subnets that
away from RIP to more sophisticated routing protocols.
are closely related. All areas must connect to the backbone area.
IGRP
OSPF provides fast rerouting and supports variable length
With the creation of the Interior Gateway Routing Protocol
subnet masks.
(IGRP) in the early 1980s, Cisco Systems was the first company
to solve the problems associated with using RIP to route data- Integrated IS-IS
grams between interior routers. IGRP determines the best path ISO 10589 (IS-IS) is an intradomain, link state, hierarchical
through an internet by examining the bandwidth and delay of routing protocol used as the DECnet Phase V routing algorithm.
the networks between routers. IGRP converges faster than RIP, It is similar in many ways to OSPF. IS-IS can operate over a
thereby avoiding the routing loops caused by disagreement over variety of subnetworks, including broadcast LANs, WANs, and
the next routing hop to be taken. Further, IGRP does not share point-to-point links.
RIP’s hop count limitation. As a result of these and other
improvements over RIP, IGRP enabled many large, complex, Integrated IS-IS is an implementation of IS-IS for more than
topologically diverse internetworks to be deployed. just OSI protocols. Today, Integrated IS-IS supports both OSI
and IP protocols.

Like all integrated routing protocols, Integrated IS-IS calls for

all routers to run a single routing algorithm. Link state adver-
tisements sent by routers running Integrated IS-IS include all
destinations running either IP or OSI network-layer protocols.
Protocols such as ARP and ICMP for IP and End System-to-
Intermediate System (ES-IS) for OSI must still be supported
by routers running Integrated IS-IS.
Exterior Routing Protocols Cisco’s TCP/IP Implementation
EGPs provide routing between autonomous systems. The two In addition to IP and TCP, the Cisco TCP/IP implementation sup-
most popular EGPs in the TCP/IP community are discussed in ports ARP, RARP, ICMP, Proxy ARP (in which the router acts as
this section. an ARP server on behalf of another device), Echo, Discard, and
Probe (an address resolution protocol developed by Hewlett-
EGP Packard Company and used on IEEE 802.3 networks). Cisco
The first widespread exterior routing protocol was the Exterior routers also can be configured to use the Domain Name System
Gateway Protocol. EGP provides dynamic connectivity but (DNS) when host name-to-address mappings are needed.
assumes that all autonomous systems are connected in a tree
topology. This was true in the early Internet but is no longer true. IP hosts need to know how to reach a router. There are several
ways in which this can be done:
Although EGP is a dynamic routing protocol, it uses a very simple
• Adding a static route in the host pointing to a router
design. It does not use metrics and therefore cannot make true
intelligent routing decisions. EGP routing updates contain net- • Running RIP or some other IGP on the host
work reachability information. In other words, they specify that
• Running the ICMP Router Discovery Protocol (IRDP)
certain networks are reachable through certain routers. Because
in the host
of its limitations with regard to today’s complex internetworks,
EGP is being phased out in favor of routing protocols such • Running Proxy ARP on the router.
as BGP.
Cisco routers support all of these methods.
BGP
BGP represents an attempt to address the most serious of EGP’s Cisco provides many TCP/IP value-added features that enhance
problems. Like EGP, BGP is an interdomain routing protocol cre- applications availability and reduce the total cost of internetwork
ated for use in the Internet core routers. Unlike EGP, BGP was ownership. The most important of these features are described
designed to prevent routing loops in arbitrary topologies and to in the following section.
allow policy-based route selection.
Access Restrictions
BGP was co-authored by a Cisco founder, and Cisco continues to Most networks have reasonably straightforward access require-
be very involved in BGP development. The latest revision of BGP, ments. To address these issues, Cisco implements access lists,
BGP4, was designed to handle the scaling problems of the grow- a scheme that prevents certain packets from entering or leaving
ing Internet. particular networks.

An access list is a sequential list of instructions to either permit

or deny access through a router interface based on IP address
or other criteria. For example, an access list could be created to
deny access to a particular resource from all computers on one
network segment but permit access from all other segments.
Another access list could be used to permit TCP connections from
any host on a local segment to any host in the Internet but to
deny all connections from the Internet into the local net except
for electronic mail connections to a particular designated mail
host. Access lists are extremely flexible, powerful security mea-
sures and are available not only for IP, but for many other proto-
cols supported by Cisco routers.
TCP/IP

Other access restrictions are provided by the Department of Suppressing Network Information
Defense-specified security extensions to IP. Cisco supports both In some cases, it may be useful to suppress information about
the Basic and the Extended security options as described in RFC certain networks. Cisco routers provide an extensive set of
1108 of the IP Security Option (IPSO). Support of both access configuration options that allow an administrator to tailor the
lists and the IPSO makes Cisco a good choice for networks where exchange of routing information within a particular routing pro-
security is an issue. tocol. Both incoming and outgoing information can be controlled
using a set of commands designed for this purpose. For example,
Tunneling networks can be excluded from routing advertisements, routing
Cisco’s TCP/IP implementation includes several schemes that updates can be prevented from reaching certain networks, and
allow foreign protocols to be tunneled through an IP network. other similar actions can be taken.
Tunneling allows network administrators to extend the size of
AppleTalk and Novell IPX networks beyond the size that their Administrative Distance
native protocols can handle. In large networks, some routers and routing protocols are more
reliable sources of routing information than others. Cisco IP
IP Multicast routing software permits the reliability of information sources to
The applications that use the TCP/IP protocol suite continue to be quantified by the network administrator with the administra-
evolve. The next set of applications will include those that use tive distance metric. When administrative distance is specified,
video and audio information. Cisco is actively involved with the the router can select between sources of routing information
Internet Engineering Task Force (IETF) in defining standards based on the reliability of the source. For example, if a router
that will enable network administrators to add audio and video uses both IGRP and RIP, one might set the administrative dis-
applications to their existing networks. Cisco will support the tances to reflect greater confidence in the IGRP information. The
Protocol Independent Multicast (PIM) standard. In addition, router would then use IGRP information when available. If the
Cisco’s implementation will provide bandwidth management, source of IGRP information failed, the router automatically
security and interoperability with the MBONE, a research multi- would use RIP information as a backup until the IGRP source
cast backbone that already exists today. became available again.
IP multicasting (the ability to send IP datagrams to multiple
nodes in a logical group) is an important building block for appli-
cations such as video. Video teleconferencing, for example,
requires the ability to send video information to multiple tele-
conference sites. If one IP multicast datagram containing video
information can be sent to multiple teleconference sites, net-
work bandwidth is saved and time synchronization is closer
to optimal.
Routing Protocol Redistribution • The current state of the active routing protocol process,
Translation between two environments using different routing including its update interval, metric weights (if applicable),
protocols requires that routes generated by one protocol be active networks for which the routing process is functioning,
redistributed into the second routing protocol environment. and routing information sources
Route redistribution gives a company the ability to run different
• The active accounting database, including the number of
routing protocols in workgroups or areas where each is particu-
packets and bytes exchanged between particular sources
larly effective. By not restricting customers to using only a single
and destinations
routing protocol, Cisco’s route redistribution feature minimizes
cost while maximizing technical advantage through diversity. • The contents of the IP cache, including the destination IP
address, the interface through which that destination is
Cisco permits routing protocol redistribution between any of its reached, the encapsulation method used, and the hardware
supported routing protocols. Static route information can also be address found at that destination
redistributed. Further, defaults can be assigned so that one rout-
ing protocol can use the same metric for all redistributed routes, • IP-related interface parameters, including whether the inter-
thereby simplifying the routing redistribution mechanism. face and interface physical layer hardware are up, whether
certain protocols (such as ICMP and Proxy ARP) are enabled,
Serverless Network Support and the current security level
Cisco pioneered the mechanisms that allow network administra-
• IP-related protocol statistics, including the number of packets
tors to build serverless networks. Helper addresses, RARP, and
BOOTP allow network administrators to place servers far away and number of errors received and sent by the following proto-
from the workstations that depend on them, thereby easing cols: IP, TCP, User Datagram Protocol (UDP), EGP, IGRP,
network design constraints. Enhanced IGRP, OSPF, IS-IS, ARP, and Probe
• Logging of all BGP, EGP, ICMP, IGRP, Enhanced IGRP, OSPF,
Network Monitoring and Debugging
IS-IS, RIP, TCP, and UDP transactions
With today’s complex, diverse network topologies, a router’s
ability to aid the monitoring and debugging process is critical. As • The number of intermediate hops taken as a packet traverses
the junction point for multiple segments, a router sees more of the network
the complete network than most other devices. Many problems • Reachability information between nodes
can be detected and/or solved using information that routinely
passes through the router. Summary
IP is one of over 20 protocols that can be simultaneously routed
The Cisco IP routing implementation provides commands that
and bridged by any of Cisco’s routers. Cisco has added features
display the following:
to its IP implementation that optimize the performance of Cisco
• The current state of the routing table, including the routing routers in larger, enterprise-wide internetworks.
protocol that derived the route, the reliability of the source,
the next IP address to send to, the router interface to use,
whether the network is subnetted, whether the network in
question is directly connected, and any routing metrics
 Corporate Headquarters European Headquarters Intercontinental Japanese Headquarters Cisco Systems has over 90
Cisco Systems, Inc. Cisco Systems Europe, Headquarters Nihon Cisco Systems K.K. sales offices worldwide. Call
170 West Tasman Drive s.a.r.l. Asia, Australia, Canada, Latin Seito Kaikan 4F, 5, Sanbancho 408 526-4000 to contact your
San Jose, CA 95134-1706 16 Avenue du Quebec America, Mexico, New Zeland Chiyoda-ku, Tokyo 102 local account representative
USA Batiment L2 Cisco Systems, Inc. Japan or, in North America, call
Tel: 408 526-4000 ZA de Courteboeuf 170 West Tasman Drive Tel: 81 3 5211 2800 800-553-NETS (6387).
800 553-NETS 91961 Les Ulis Cedex, San Jose, CA 95134-1706 Fax: 81 3 5211 2810
(6387) France USA
Fax: 408 526-4100 Tel: 33 1 6918 6100 Tel: 408 526-7660
Fax: 33 1 6928 8326 408 526-4646

European Offices
Belgium Norway Sweden Asia Taiwan Office 150 King Street West New Zealand
Cisco Systems Belgium, Cisco Systems Cisco Systems Cisco Systems Hong Kong, Formosa Business Center, 3F Suite 1707 Cisco Systems New Zealand
Complex Antares Holmens gate 4 Stockholms Modecenter Ltd. 285 Nanking East Road Section Toronto, Ontario M5H 1J9 Level 16, ASB Bank Centre
Avenue des Pleiades 71 0250 Oslo, Norway S-117 60 Stockholm, Suite 2704 3 Canada 135 Albert Street
1200 Brussels, Belgium Tel: 47 22 83 06 31 Sweden Far East Finance Center Taipei, Taiwan, R.O.C. Tel: 416 506-1500 P.O. Box 6624
Tel: 32 2 643 2626 Fax: 47 22 83 22 12 Tel: 46 8 681 41 60 No. 16 Harcourt Road Tel: 86 2 719 8383 Fax: 416 506-1506 Auckland, New Zealand
Fax: 32 2 643 2627 Fax: 46 8 19 04 24 Hong Kong Fax: 886 2 719 4537 Tel: 64 9 358 3776
Spain Tel: 852 529 3534 Latin America Fax: 64 9 358 4442
Germany Cisco Systems Spain Switzerland Fax: 852 520 2676 Cisco Systems, Inc.
Cisco Systems GmbH Paseo de la Castellana 141 Cisco Systems Switzerland Australia 170 West Tasman Drive
Max–Planck Strasse 7 pl.18 Edificio Cuzco IV Sonnenberg 5 Cisco Systems (HK) Ltd. Cisco Systems Australia San Jose, CA 95134-1706
85716 Unterschleissheim, 28046 Madrid, Spain 8636 Wald, Switzerland Shell Tower, Level 37 Pty., Ltd. Tel: 408 526-7660
Germany Tel: 34 1 572 0360 Tel: 41 55 95 60 44 50 Raffles Place Level 17, 99 Walker Street Fax: 408 526-4646
Tel: 49 89 3215 070 Fax: 34 1 570 4599 Fax: 41 55 95 64 14 Singapore 0104 P.O. Box 469
Fax: 49 89 3215 0710 Tel: 65 320 8350 North Sydney, NSW 2060 Mexico
United Kingdom Fax: 65 320 8307 Australia Cisco Systems de México,
Italy Cisco Systems Ltd., Unit 3 Tel: 61 2 957 4944 S.A. de C.V.
Cisco Systems Italy Cliveden Office Village Cisco Systems (HK) Ltd. Fax: 61 2 957 4077 “Torre Caballito”
Office No 609, 6th Floor Lancaster Road Paso de la Reforma
Via Turati 28 High Wycombe, Canada No. 10,, Piso 19
20121 Milan, Italy Bucks HP12 3YZ Cisco Systems Canada Colonia Centro
Tel: 39 2 62 726 43 United Kingdom Limited México, D.F. 06400
Fax: 39 2 62 729 13 Tel: 44 494 464944 Tel: 525 628 1255
Fax: 44 494 465300 Fax: 525 628 1212

Copyright  1994 Cisco Systems, Inc. All rights reserved. Printed in USA.Cisco Systems and the Cisco logo are registered trademarks of Cisco Systems, Inc. All other products or services trademarks, service marks, registered trademarks, or registered service marks
mentioned in this document are the property of their respective owners. Lit. # 010103 4/94 GP