Part I: Background

Chapter I presents an overview of tel.ecommunicaiions, dala communlCIItions, and network

management. It is a broad review of networking and network mamigement. It stru1s with an anabgy
of lite telephone network. Telephone network almost always works, and there are reasons tOr its
achieving quality and reliability. You will learn lite relationship between data communications and
telecommunications and b.ow the d.istinction between the two is slowly disappearing. The influence
of desktop computing sod distrlbuted computing environment based on client-server architecture
.b as revolutionized computer communlCIItion. The lnier.n et is a worldwide fabric and you willleam
to appreciate bow infOrmation travels across it around the globe. Basics of communication protocols
and architecture are presenred along with various standard~. Select equivalent applications are used
as illustrations comparing the Internet and OSI protocols.

Components of net work llUIJl8gemeot are described and complemented by interviews witll net work
managers, whose experiences emphasize the need .fur network managemem and a network
operations renter. Network management is more than just managing networks. Network
management l$ presented from lite perspectives of service management, operatiorts support systems,
and business management. The platfOrm !Or a network management system is discussed based on
client- server architecture. Chapter I concludes with a note on future trends in network management
technology.

Chapter 2 !Oc:uses on net\vork techoology. You may skip this chapter if you are familiar with the
practical aspects of networking. If you are knowledgeable on principles of data communication, this
chapter. will help you appreciate lite 1eclloological aspects of i1. You will learn how various
topologies are implemented in LAN and WAN networks. Basics of Ethernet, Token Ring, and
FOOl networks are described from a pmctical point of view. Ofll"oe.se, Elhernet is the most widely
deployed LAN today. LAN evolution from basic Ethernet to Gigabit Ethernet. with half· and full·
duplex configurations is presented. Switched Ethernet adds capability to expand lite bandwidth and
the flexibility of LAN. Virtual LAN is implemented using a switched Ellternet bub accomplishing
flexibility in administmtion of workstations across mukiple LANs. You will learn the various
network components--hubs, bridges, routers, gateways, and protocol converters--that need to be
managed. A brief review of wide area networking and transmission technology is also presented.
Broadband technology is briefly described in Litis chapter. but a dem.iled discussion of it will be
done in Pan rv while addressing the managemem of broadband networks and services.
1. Data Communications and Ne twork Ma nagement Overview

Objccth'es

Teleoommunications overview
Data oom.municalions overview
Evolution ofconverged networks
Desktop processors and LAN technology
Client-Server architec1ure in networking
Internet and intmnet
Nerwork communication protocols
OST and lntemet standards
Broadband networks and services
Need for network management and NMS
Operat·ions, Administmtion, Ma.inlenance, and Provisioning
Network management architecture and organization
Concep1 ofNetwork Operations Center
Perspeclivc.s of network managemenl
• Network ma11agemeli/ :r."-">tem
Look-ahead of network management technology

This chapter demonstrates the lleCessity of network system and service management in providing infurmation
technology (IT) services. The challenges thai IT managers face are pl:esented to motivate !he studenl to get excited
about n.etwork management. We start with the ltistory of computer oommunicatiotL walk you through some real-
world case histories, and then present an overview of various aspecls of network management.

The telephone system is known to be very reliable and dependable, One can make a lelepbone call from anywhere
to anywhere at any time oft.be day and be reasonably sure that the connection will be made and the quality of
conneclion will be good. This is partly due lo the efficient management of the telephone network. Secrion 1.1
introduces the concept of managemem for the success of telephone nelwork by using Operation Support Systems
(OSSs).

Computer communication initially used the 'telephone networlt to carry digital data. There was a clear demarcation
between '!be tmditionalteleconummication network and computer communication network. The evolution of early
computer communication networks is dealt with in Section 1.2.

Computer communicalion technology radically changed with the advent of desktop computing power and
distributed computing environments (DCEs) using local area net\vorks (LAN) as described in Sec! ion 1.3. Global
communicalion using Internet became. a reality with the introduction of TCP/IP-based networks. Section 1.4
describes Internet and inlnlnel followed by a dlscussion in Section 1.5 on the importance of communication
pro!oonls and standards.

The nexi phase in the evolution of IT was !he introduerion of broadband services. Voice, video, and data oould he
delivered on the same medium to homes. lltis has revolutionized the access network to horne and the distribution
network at customer premises. 11 has also iniliated impt"Ovemenl in the core wide area network (WAN). Section 1.6
addresses these issues.

Nelworking is full of " war stories" as experienced by IT managers. Sec! ions I.7 and 1.8 presenl case histories
experienced by IT managers and the challenges !hey face in today' s computer and telecommunicalion
enviro.mnent. Interviews with them emphasize the impor11loce of network and system management tools. Section
1.9 describes network management that comprises operations, administration, maimenan.ce, aod provisioning.
Three groups perform these functions: Engineering, Operations, and Installation aod Maintenance (l&M). Section
I. 10 focuses on Network Management System (NMS) aod relationships between its various components. Besides
managing network components, application system resources also need to be managed. TWs is the subject of
Section 1.1 I.

Network. managemem tecbnology is still in an evolutionary mode as network and software technologies advance.
Section 1.12 bricily add.resses NMS platforms baSed on Microsoft. Windows and UNIX operating system. The
future direction~ of network management technology form the content of Section L 13. As with all chapters in the
book. a sun:unary section and exercises conclude this chapter.

1.1. Analogy ofTclcph on e Networ k Management

The need. for data or computer communication network management is best illustrated. by an analogy of telephone
·network management. Tbe high degree of reliability of the telephone network is evidenced by the following
illustration. We can pick up a telephone, call anybody, anytime. anywhere in tl!e world, aod be· almost sure to be
connected to the destination. It is reliable and dependable; and the quality ru1d speed of connection are good. It is
reliable because it almost a !ways provides se.rvice of voice communication tbat we expect of it. It is dependable
because we can be fairly sure that il works when we need il. especially in an e.mergency situation. such as 911 calls
in 'the USA or military defense situations. The quality of service is generally good; and we can have a conversation
across !he world with the same clarity that we bave when we call our neighbor.

The pre~nt-dny telephone network is referred to as Public-Switched Telephone Network (PSTN). aod is probably
the best example of traffic engineering providing guaranteed Quality of Service. The reason .for such reliabil ity,
dependability, and quality Is more than careful planning. design. aod implementation of a good telephone network
using good and reliable components. 'Tlte key is management aod operation of the oetwor.k. Much of the
management of the network is so well automated that it becomes part of the operation. Let us first look at the
telephone net work architecture and then at some oflhe operations support systems tbat manage it. In the 1970s the
telecommunications industJy switched to digital services, which followed much the same pattern as voice services
and conceived a vision of end-to-end circuil-switc.hed services. known as the Broadband Integrated Services
Digital Network (B-ISON). B-ISON is now being replaced by Internet and Broadband Service.

The ar.chitecntre of a telephone network is hierarchical as shown in Figure 1. 1 [AT&T 1977]. There are five levels
of network switches and three types of trunks that connect these switches. A trunk is a logical link between two
switches and may traverse one or more physical links. The end office (Class 5), wh.ich is lowest in the hierarchy, ls
the local switching office. The customer's telephone or Private Branch Exchange (PBX) Is connected to the end
office via a dedicated link called " loop." The other fuur higher levels of switches (Class 4 through Class I) are
tandem or toll switches carrying toll (long-<listance) calls. Because of the advance in switching technology and
economy of transmission, ctasse·s I through 4 have been merged into a single class rercrred to as Class 4. A direct
trunk connects two end offices, a toll-connecting trunk connects an end office to any toll office, and 11.toll (internal)
trunk connects any two toll offices.

Figure 1.1. Td•t•hou• Notwork Mod<l

 To Oliler
R egional Centers
Sedional Cenlers
Primal)' Centers
ToUCenteB
EndOff.a;s

Primary Centers
Toft Centers
EndOiftee$

Cia$$ 4 Toll Polnls

EndOtr,._

End Cllfoee EndOfflc<>

Class 5 SwiiCh Class 5 Switch
Logond;

• t -
.... I..OQP
Olrecl Trunk

6> Voire
0 Voice
-
-
- Toii.COnnoc~ng Trunk
Toii Trunk

From the lo<:al Class 5 office to the called party's Class 5 office, there are multiple routes. A circuit connection is
set. up either directly using a local trunk or via higher-level switches and routers. Primary and secondary routes are
alrea~y programmed into the switc.h. lftbe primary route Is broken or facilities over the primary route are filled to
capacity, an alternate route is automatically assigned. For e.xrunple, on Mother's Day, which is the busiest
telephone-troffic day oft he year in the United States, a call io tbe neighboring town could trove! clear across the
country and. back iftb.at 's the route where adequate bandwidth is available. Let us remember that there is a 3-hour
time difference· between the two coasts, and traffic in the West Coast starts 3 hours laterlhan the East Coast.

To ensure tbe quality of service in a telephone network, o perations support systems are implemented. They
constantly monitor the various parometerll oftbe network. Forexa.mpte, to ensure that there is adequate band,vidLb
to carry the· troffic over the facilities, a troffic measurement system constantly measures iroffic over switch
appearances. The results are analyzed for facility-planning purposes. They also provide real-time input to a NMS
when there is excessive blocking (traffic over the capacity oft.he trunk group) in any link.

The qua lily of the call, measured in terms of signal-t~no ise (SIN) rotio, is measured regularly by a trunk
maintenance. system. This system accesses all the trunks in an office during the night· and does a loo~back test to
the far end. lbe results are analyz.ed in the morning and corrective actions taken. For example, if the SIN ral.io of a
trunk is below the ac.ceplaoce level, the trunk is nimovcd from service before the· customer experiel.lCes poor
performance.
Par a given region, there is a network operntions cemer (NOC) where the global s1atus of !he network is mo.nito red.
Traffic patterns are constant.ly observed and corrective operations are taken, if needed, in real time. The NOC is t:he
nerve center of telephone network operations.

lt is worlh noting !hat the telephone .network is lll1IJI8ged from the users' perspective, and not nom that of the
system or the service provider, even though the objectives of both are the same. However, with emphasis on the
user's point ofv.iew, the fnt objective in operations is restoration of service and then the qualiiy and economy of
service. Thus, isolation of t he problem and providing allelll8tive means of service, by either manual or automated
means, bocome more important !han fixing tbe problem.

To manage a network remotely, i.e., to monitor and control network componems from a central location. network
management functions need to he bui.lt into the components of the network as much as possible. In tbat sense,
network component designs should include network 01anagement functions as part of their requirements and
speci.fications.

The computer or da.ta communication network has no1 m3tured to the same ell1em a.s the telephone network. Data
communications technology is merging with telephone technology. Dala and modern telecommunication networks
are evolving into broadband communication networks and are more complicated 1han the plain old telephone.
service (POTS). Analog audio and video services are migrating to digillll services. The analog hierarchy of low-to-
high bandwidth signals is being mmsmitted across the globe using a Synchronous Digital Hierarchy (SOH) mode.

Network management and operations of lhese digital networks are continuously being developed as new
l.echno logies emerge. Furlher, the telephone industry all over the world had been monopolistic and 1hus singlt>
vendor oriented. Thi s is no longer true. Digital-based computer communications started 8ll a private industry and iS
hence m1~tivendor oriented. Unfortunately, this bas produced enoanous problems 10 users because network
components supplied by different vendors do not always communicate with eacb other. The network or
information systems manager. who has the responsibility of keeping the service alive all the time, has been
confronted with resolving the issue as oow tecbnology and oow vendor products emanate. This situation has been
recognized by various industrial and standard groups and is being continuously addressed.

1.2. Da.hl (Computer) 110d T elecomm unication Network

Network communications lechnology deals with the lheorY and application of electrical engiooering, computer
engineering. and computer science to aU types of communication over networks. It also addresses accessing of
daUlbases and applications remotely over LANs 8ll well as switched and private lines. A basic network can be
vJe,ved as interconnected nodes and links as shown in Figure 1.2. A link carries infurmalion from one node to
another that. is directly connected to it. A node behaves as an end (tenninating or originaling) node, or an
intennediate node. or both. If the node behaves as an end node, infonnation either originates or tenninates there.
An intermediate node redirects the informal ion from one link to another. End-office nodes mentioned in Section
1.1 behave as end nodes. A node can drop and add infur01ation channels and til the same time switch infurmation
t·ransparently between two links. EaCh end node has a connection to a user interface if1he infOrmation originates or
termina1es !here. Tlti.s interface could use My type of equipment-audio, video, or Data Tenninating Equipment
(DTB). A DTE is any equipment I hat generates or accepts digital da1a.

Flgul't 1.2. LogicAl Nttwnrk Modtt

 VIdeo

EN: End NOC!o

IN. lnlermodialo NIXIe Wo.rkslallon

Daia coo be trnnsmilled eiiher in an analog or digital formal. The analog da!a are sent eilher·as a baseband (e.g.,
voice data from !he switching office 10 !he customer premises) or on top of a carrie.r (e.g., cable TV). Digital data
are eilher directly generated by the user equipmem (e.g., computer 1erminnl) or as analog data and are convened lo
digital data (e.g., Integrated Services Digital Network. (ISDN) connection to cus1omer premises). The ]utter
scenario of the ability to handle integrated digital and analog signals is beooming extremely importanl as in the
case of multimedia broadband services. Management considerations associated with them are alsO very
oballenging, as we will sec in Pan lV. Long.<fistance data transmission today is mos1ly digital due 10 its superior
prioe and performance.

Data ore sen! from the originating to the terminating node via a direct link. or via a tandem of liok.s and intermediate
nodes. Dala can be transmiued in one of three modes: circuit switched, message switched, or packet switched. Ln
!he circuit-switched mode, a physical circuit is established between the originating and terminating ends befure the
data are transmitted. The circuit is released or "tom down" after completion of transmission.

ln message-switched and packet-switched modes, data are broken int·o packets and each packet is enveloped with
destination and originating addresSes. 1lte message-switched mode is used to send long messages) such as emall.
The packet-switched mode is used to transmit small packets used in applications such as intera~1ive
communication. Bridges and routers open each packet to find the destination addre$5 and switch the data to the
appropriate output links. The path between the two ends may change during the transmission of a message because
each packel may take a differenl route. They are reassembled in I he right order at the receiving end. The main
difference between message and packet swilcb.ing is thm in the former, data are stored by !he system and then
retrieved by the user at a later time (e.g, email). l,o the packet-switched mode, packets fire fragmented and
reassembled in almost real t·ime. They are stored in the system only long enough to receive aU the packets in the
message. ln Europe, X.25 packet-switched network was ex-tensively used in Public-Switched Data Network
(PSDN).

Neh~rk communications are commonly classified as either data communications or telecommunications. This
classiftcation is based on historical evolution. The te.lephone network, which came into existence first, was known
as a telecommunication network. It is a circuir-swilcbed network that is structured as a public network accessible
by any user. The telephone network represents a telecommunication net,vork. The org-anization that provides this
service is called a telecommunication service provider (e.g., AT&T, British Telecom, NTI, BSNL, etc.).

With the advent ofoomptrtcrs, the terminology data communication network carne into vogue. It is also sometimes
called computer communication network. The telecommun1catiorts infrastructure was, and is, still used for data
communications. Figure 1.3 shows an early configuration of terminal-to-host and host-to-host commuoicat.ions,
and how data and telecommunication net works interface w ilh each other. To interface, a terminal or host connected
to an end-office switch communicates with the host connected to another end-office switch by modems at each
end. Modems transfer information from digitallo analog ai the source (telephone nehvorks carried analog signals)
and back to digilal at the destinatk>n.

Figurt t.l.A notog•nd Data Ttltrommunlu rlon Nttwor ks

Modem telecommunication networks mostly carry digital data. The nodes in Figure 1.4 are digital switches.
Analog signals from telephones are converted to digital signals either at the customer premises or the central office.
figure 1.4 shows a corporate or enterprise .environment in the stage of the .evolution of data a.nd telephone
communications. A number of telephones and computer terminals at various corporate sites are connected by
telecommunication network. Telephones are locally interconne<:ted to each other by a local switch, PBX, at the
customer premises, which interfaces digitally to the telephone .network. The computer terminals arecoMected to a
communication controller, such as a djgital mutt iplexer, which provides a single interface to the telephone network.

Figurt 1..&.. Oiaie:al Data and Te:IKOIIHUWJication Ntt\VOI'k s

With the advent of desktop computers and LAN, data communication was revolutionized. Desktop computers
could communicate with each other over the LAN. This led to a Distributed Computing Environment (DCE),
which is discussed in tbe next section.

1.3. DL~ttibuted Computing E nviron ment

Figure 1.5 shows a LAN with hosts and workstations. Let us ob.serve that dtey are workstations with processing
power and not just dumb terminals as described in the previous section. Any workstation can communicate with
any host on the LAN. There can be a large oumber ofworkstatioDS and bosts depending on the type of LAN. D1Es
connected to different LANs that are gcogrnpbically far apari can communi cate via telecommunication net,vork,
either public or private switched. The system of links eotu1ecting remote LANs is called a WAN. A LAN is
physically connected to a WAN by a bridge or a router as shown in Figure I.S(b). We will discuss the types of
LANs and WANs in Chapter 2 . .First, we want to bring out two important aspects ofDCE in this section.

Figurt t ..5. DCE with L~Ns IUid WANs

 ~ 1l g~
w...- ._ W<NWon

0 I ElJiemot I

~
I

w-..
(•1-ondVI-ool-"CCOIAo'l
1i-·
Q~
/ =
:=-'6 LANC

-%-- WNI

The first aspect is the question of whether the different plalforms and applications mooing on DCBs have the
ability to communicate with each other. In the early stage of communication network evolution, proprietary
interfaces between platforms and processeJ> were implememed by telecommunication service providers and
computer vendors to communicate autonomously within each of their networks. Fo r example, Bell System, a
monopo listic telecommunlcatbn service provider, and IDM, the largest computer vendor, established transmission,
switching, and interface standards and manufuctured their own communications equipment to meet them. They
made significant contributions to the standards bodies to make such specifications the industry standards. For
customer premises equipment (CPE) interface, specifications are published for them to interlilce cleanly with the
network. For example, Bell System published specifications for Customer Service Unit (CSU) for customer
equipment to interface with the network. However, as the telecommunications industry rapidly grew, national and
intematb nal standards needed to be established for communication between equipment provided by various
vendors. Protocols and database st11ndards for handshaking and information exchan.ge are discussed in the
following sections. For now, we will assume that the different processors 1llld prooesses running on them collld
communicate with each other.

The second aspect of DCB is the ability of processors attached to LANs to do multiple functlons. They could
continue, as dumb terminals did, to request a bost to perfunn the functions and retmn the results. Alternatively,
they could request some special functions to be performed by a host-and it could be any processor in the
network-and receive the results. In this scenario, the processor that requests a service is called the client; and the
processor tbat provides the service is called the server. Such a co nfiguration is termed a client- Saver environment.
Allhough the terminology of clieru and server is commonly associated with t.he pro cessors, the more accurate
defnition shouk! be associated with the processes. Thus, the process that iniliates a transaction 1.0 run an
application in either a local or a remote processor is called the client. The application process that is invoked by a
client process is called ·the server. 1lte .serv-er returns the results to the cl ient. Tbe application designed to take
advantage of such a capability in a network is called a clie.m- server architecture. With such an interpretation, the
e llen! and server processes can coexist in the same processor or in different processors.

We will now go into some detail on ·the salient characteristics and features of client-server architecture and models,
as they are very pertinent to nct\\~:>rk management applications and architecture. A simple client-server model is
s hown in Fig ure 1.6. There is apt to be confusion between wh.ich is a c lient and which is a server in distributed
computing architedure. The 'best way to distinguish 'between the two is to remember that the client initiat es the
request and the server responds.

Flgw·• 1.6. Simplt Clitni-Str\'tr Modtl

Cllel'\l

The client initiates a request to tbe server and waits. Tbe server executes tbe prooess to provide the requeSied
serv ice and sends the resuk$ to the client. lt is worth noting that the client cannot initiate a process in thu server.
Thus. t be p rocess should have already been started in the server and be waiting for request.s to be processed.

A rea.l-world analogy to the c l ieot-server operation is a post offioe. The clerk behind the counter is ready and
waiting for a cl ient. She is a server. When a customer walks in and initiates a transaction, for example, ordering
stamps, the clerk responds. The c ustomer is the client. After the clerk gives me Sinmps to the customer, I.e., she has
delivered the resuks, the customer leaves and the cler.k, as a server, goes into a waiting mode until the next client
initiates a transaction.

As with any system, delays and breakdowns of COIIliDuoiCation need to be considered in this model The server
may be providing the service to many clients that are conncctoo to it on a LAN. as shown in Figure I. 7(a). Eaeh
client's request is normally proces.~ed by the server according to the FrFO rule--f~rst in first out. This dei3y could
be min.im.ized, but not eliminated, by concun:eot processing of requests by the server. It is also possible that, due to
either the communication link or some other abnormal termination, the server may never return the result to th.c
client. The application on the client s hould be programmed to take care of sllCh deficiencies in communication.

Fi~ur< 1.7. Clitni-Strvt r In Discrlbused Cumt>ulio.g.Environmeuc

 g . .~. .

'
..
·
~,
..f
'\ ··....
··•., ......~.~. ... ..... ..__.......... .•'
...,-
...~. ............__ ,.,.. _.,.~·····
··.................. ________, .........................

_ ........-.....--~ - ~
........- ····
.•
..· ..··...
.~

: ...~
.. i

'\.... ·.

·..

Since the client and application are processes runnlng in nDCB, each oftbem can be designed to execute a specific
function efficiently. Further. each function may be under the jurisdiction of different departments in an
organization . An example of this is shown in Figure 1.7(b). joe.stooe@source.com (Joe Stone's user id) using a
client in a network sends a message to sally.jones@lest.com (Sa lly Jones' user id) on the network. The message
first goes to tbe mall server on the network. BefOre it can pro<:ess the request, the mall server needs to know the
netwo rk !!ddress of sally.jones, which is dest.com. Therefore, it makes a request to the domain name server (DNS)
on the network for routing information for the address of dest.com. When it receives that information, it sends out
joe.stone' s message via the bridge connected to the network. It then sends a message to joe.stone on the client
stating that the message has been sent (o r not sent because the dest.com address does not exist in the DNS). In ·this
e-xample, the mail se.rver behaves both as a server and as a client. The three processes in this scenario, namely the
client, lhc mail server, and the DNS, are considered cooperative computing processes and may be running in three
separate platfonns o n remote LANs connected by a WA:N. Communication between these processes is called peer-
to-peer communication. We wi.ll soon learn bow network. management fits into such a model nod manages
components on lbe network. that perform cooperative computing using peer-to-peer communication. However,
befOre we pursue that.. let ns first look at a new dimension that the DCE has caused ·networking to mushroom
into-t:helntemet.
lA. TCP/JP-Based Networks: Internet and .In tranet

Transmission Control Prot.ocoVlmemCI Prot.ocol (TCPIIP) Js a suile o f protocols that enable networks to be
interconnected. It forms the basic foundation oft he Internet. Architecture and protocols are discussed in detail in
Section 1.5. We will brie·fly Mscribe tbe role TCP/IP plays in Internet. Nodes in the network route packets using
network protocol IP, a connectionless protocol. That means there is no guarantee that the packet will be delivered
to the destination node. However, end-to-end communication can be guaranteed by using the transport protoco~
TCP. Thus, if a packet is lost by LP, the acknowledgement process ofTCP ensures successful retrans mission of the
packet.

TCP/TP suite of protocols contains more than TCP and IP protocols. TCP is a connection-oriented protocol A
complement to TCP is User Dalagrnm Protocol (UDP)., which is a co.nnectionless protocol. Much oflntemet traffic
really uses UDPIIP due to the reliability of data transmission. For example, email and management messages are
carried by connection less transmission.

lbe Internet is a net work of networks. Just as we can communicate over the telecommunication ·network using the
telephone from anywhere to anywhere in the world today, we can now communicate worldwide over the computer
network via email. We looked at the example of Joe St.o ne sending a message to Sally Jones in the previous
section, Figure 1.7(b). Let us expand that example and visualize that Joe Stone, who is at the College ofComptrting
building of Georgia lnstintte of Technology, is sending an email to Sally Jones at her home in Australia. SaJiy is
connected to an Internet service provider, ostrich. com. Similar to a unique telephone number that each station has
in the telephone world, each person has a unique address in the computer communication netwOrk. Joe's email
address isjoe@cc.gatecb.edu and SaUy's address issally@ostrlcb.com.au.

Figure 1.8 shows an lnt.emet configuration fOr our scenario. Assume that Joe is at Workstation A on LAN A
sending the ernall to Sally at Workstation Z that is "telccormected" to her Internet service provider's email server
on LAN Z. Two serve.r s shown on LA~ A are mail server and DNS. J't should be noted that the servers do not b.we
to be on the same LAN as the sender's LAN, as shown in Figure 1.8. The two servers cooperatively transmit the
email message to LANCon the computer network made up of bridges and routers. lbe link between LAN A and
LAN C could be a WAN. tnfurmation is transported exclusively based on TCPITP-based protocols. We will explain
TCP/LP protocol in Section 1.5.2.

Flgurt t .8. 1nttrnet Conllgurnclon

lnfonnation from LAN C progresses vill gateways and WANs to the computer communications network in
Australia, as shown in Pigure 1.8. The WAN network shown is composed of a ser.i es of networks, not all
necessarily using TCP/TP prot~ol Gateways between them serve as the interfaces between dissimilar and
independent autonomous networks and perfurm many functions including protocol COilversions. Autnnomous
networks have liitle knowledge of each other's aitrlbutes, configumlions, and addresses and yet communication is
automatically taken care ofey a bierarcbyoflnternet servers along the path.

Joe's email message finally reaches the email server on LAN Z in Australia and is stored there until Sally retrieves
it via her Internet link with an Internet service provider's server. ln fact, email messages are trMsmitie.d by a
" store-and-furward" scheme all along lhe path. 1n addition, the final sLage in the Lntemetlink uses a TCPIIP suite
of protocols.

Thus, via the Interne!, any user can communicate with any other user in any pan of the world as IQng as both are
connected to n network that is pan of the lnternel. This .has a Iso revolutionized the software user interfuce
providing capabilities li.ke web pages so that you can gather information about any1hing in the world instantly
through the 1nternet.

Another per.speclive of the Internet is to view it as a layered architecture, as shown in Figure 1.9. This architecture
shows the global Internet as concentric layers of workstations, LANs, and WANs interconnected by fubries of
Medium Access Controls (MACs), switches, and gateways. Workstations belong to the user plane, LANs to the
LAN plane. and WANs to the WAN plane. The interlilces are defined as the fabdcs. MAC fabric interfaces the
user plane t·o the LAN plane. LAN and WAN planes interface through switching fabric. WANs in tlte WAN plane
interface wii.h each other via the gateway fa'bric.

Flgun 1.9. loltrott Fobri< Mod<l

USER PLANE

~~ GatowiiY IOb<icl

( ] SWitchlno lab<lc
MACiob1C

The user' s workstation intermces to a LAN via a MAC. which will be explained in Chapter 2. LANs internee to a
WAN by a switching fabric of bridges, routers, and switches. Enoh WAN may be considered as an acrtooomotl~
network, and hence needs a gateway to communicate with another WAN. Gateway fllbric iotercOtmects different
WANs. Thus, a single lnternet plane at the core of the model multiplies into millions and mlllions of users at the
user plane, with virtuaiJy no limits in sight.

Communlcation between two users in the user plane, i.e., logical link connection on the user plane, takes the
following path. The physical path traverses the MAC fabric, the LAN plane, the switching fabric, the WAN plane,
and the gateway fabric to the core and then returns (o the user plane going through all the planes and interface
fabrics in reverse..

The huge success of lnternet teclmology has spawned Intranet h::chnology. The main distin~1 ion between lhe two is
s imilar to that between public and private switched networks. An intranet is a private network and access to it is
controlled by the enterprise that owns it. whereas the Internet is public.

The impaot of ihe Internet in nel\vorking is enormous. How do we manage the Inte.met? For example, if an email
does not reach its destination, how do we detect where the communication broke down? How do we take
advan111ge of lnten-.et ·capabilities to impl.eme.nt network manage.ment? We have not yet defined network
management and how it fits into the client-server environment. However, before we define what network
management is, let us briefly look at the protocols and protocol architecture that enable successful communication
between different components on the network.

1.5. Commun.ic:ttion Protocols and Standar ds

Consider a fax machine and a modem bought from a local store successfully sending a fux to a modem and fax
machine anywhere in the wodd. even though each fax machine and attacl-.ed modem were manufuctured by local
vendors. Likewise, isn't it a technological miracle thatt\\OJ computers located anywhere in the world can transmit
messages to each other as long as each is connected to the Internet? The key to the practical success of these and
other such teohnologies is the interoperability of ihe two end devices. More and more vendors in more and more.
countries have recognized that in this world of shrinking cyberspace and advancing modem communication
technology, interoperability is the key to the success of their business,

Universal interoperabllity is achieved when all participants agree to estllblish common operational procedures. In
communications lingo, commonal.lty can be interpreted as standards and procedures as protocols. Let us consider
the scenario of Joe sending an email from Georgia Institute ofTechnology(GA Tech) in Atlanta to a colleague in a
Japanese Telecommunications Company (ITC) in Tokyo. Joe oomposes the message on hi s comprrter terminal and
sends it to llis colleague (yoho@jtc.com.jp). Joe's message with his user id (joe@cc.gatech.edu) and IP address
(169. I 11.103.44) goes through several changes befure it is transmitted on tl-.e plrysical LAN medium at GA Tech.
The message goes to its College of Computing (cc)'s email server, which oblnins the IP address of the destination
and sends the message out on the Internet. The message traverses several nodes and links and arrives at the post
office box ofYoho's mail server at JTC. She establishes a session in her computer and gets the complete message
thai Joe transmitted. ln this seenario, Joe's message is wrapped with several layers of control information at
various times and is broken down into packet unitS and reassembled at the destination. A II these steps happen each
time without any loss or error in ihe message due to standard.ization and modular (layered) architecture of data
communication protocols. As we w Wsoon learn in this section, the popularity oflnternet as a peer-to- peer network
has been mnde possible by the peer-to-peer protocol TCP/IP suite.

Architecture can be defirled as ·modeling a system into functional co mponents and the relationship among ihem.
Thus, communication architecture deseribc.s the functional components of oommunication network as well as the·
operational intcr:filce between dtem. Operational procedures-both intra· and iuter-modules--ere specified in te.rms
of protocols. Just as human communication is made mutually understandable by speaking a common language,
communication protocols are standardized for service interfaces from the perspectives of both a service provider
and a service user. If diffi:rent vendors implement the same standards in thei.r sy·stem components, then
communic11tion between iheir different components can be universal. Standa.rdization of protocols involves
agreement in the physical characteristics and operational procedures between communication equipment providing
similar functions. Thus. looking at our example, all fax machines are able to comrnunicale with eacb other because
 all vendors bave implemented standards recommended by loternational Telecommun.ication Union-
Telecommunications Sec10r (ITU-T). Similarly, email exchange across !he world ls possible because most vendors
have adopted I ntemet standard Simple Mai l Transport Pm10eol (SMTP) in ·!heir software. However, there are email
software packages other than SMTP, and the user has to i.os.tall a gateway in those. systems to convert back and
filrlh between SMTP and the vendor-specific proprietary protocol For example, lBM Lotus uses oc:mail (now
defunct), and any network tl111t uses oc:mail bas to implement a gateway (o send an e.mail over the Internet. Note
·!hat there are different mail protocols (SMTP, !MAP, POP, etc.~ whlcb have different procedures. We will now
look atlhe details of communication architecture.

1.5.1. Comm m1ication A•·ch itectures

Communication between users (human beings using a system) and applications (programs that run in a system)
occurs at various levels. They can communicate wah each olher at the application level, lhe highest level of
communication architecture. Alternatively, they can exchange Information at the lowest level, the physical
mediUl.D. Each system can be broadly subdivided into two sets of communication layers. The top set of layers
consists of application layers and the bottom ~1 tronsport layers. The users-nd users include application
program!O-interface with the application level layer, and the communication equipment interfaces with the
physical medium. The basic communication architecture 1$ s.hown in Flgure 1.10. In Figure 1.1 O(a), the two end
systems associated with the two end nodes communicat.e dirCclty with eaob other. Direct communication occurs
between lhe corresponding cooperating layers of each system. Thus, lransport layers can exchange infonnatioo
with each other, and so can lhe application layers and the users.

Sytto,.z

t•l OOiet CoMmunl<a!M BMw..,. EI\C!Sy<l...,.

SVaollnA '"*"'-••)'11om
[ uwA J
}ljlpiOOim~-- ~001\ l.t)U/1

- -
Tr•l\llpol1 la)'llt
lflnOI>GrlL"Y',.
-
CGftWtc.Mwt
T_Loy. .

I Pllytlc:aii.IO<II'--1 I I Pl1l<oea1Moclillm I
This can be illustrated w.itb a real-life example. A bearing-impaired pers~n, accompanied by an interpreter,
arteoded one of my classes. As llecmred, the interpreter t.ranslated to the student using sign .language. If the student
had a que·stion, the interpreter translated the information &om sign language, orally to the da.~s and me. In this
llluslration, the bearing-impaired Sll1dent aod I are at the application layer. The interpreter di:l the protocol
conversion attheapplication layer level. 1lte transport layer is the aural aod vL~al media .

Figure I.J ()(b) shows the end systems communicating v.ia an intermediate system N, which enables the use of
different physical media for the two end systems. System N converts the transport layer information into the
appropriate protocols. Thus, system A could be on a copper wire LAN and system Z co uld be on a fiber optic
cable.

Var.ious standard organizations propose, deliberate, and establish standards. One of the internationally re·nowned
standard organizatiolls is International Standards Organization (ISO). lSO has developed a highly modular, o r
layered, architecture for communication protocols that is called the Open Systems Interconnection (OSI) Reference
Model, published as OSl RM-ISO 7498. This model was developed based on the premise abat the different layers
of protocol provide different services; and tbat each layer can communicate with only its own neighboring level.
Two systems can communicate on a peer-to-peer level, Lbat is. at the same level ofthe protocol. The OSl protocol
architecture with all seven layers is shown in Figure 1.11. Table 1.1 describes the salient features of, and services
provided by, each layer. Layers 1-4 are the transpo rt system protocol layers and layers 5-7 aro application support
protocol layers.

Figur t 1. 11. 051 Protocol Laytrs

l~7
t
Aj>f>li<ooon

layet& Pr~

"-5 ~"

~·

-
TIMspo<t

""""'
~2 0."'-"'k

L#)'tlrl f'nyu:al

T ablt 1. 1. OS I Lay•l"5 and Strvi<ts

Layer Layer Name· Salient Services Provided by the·Layer
No.

1 Physical -Transfers to and gathers from the physical medium raw bit data
Tabl t 1.1.. OSI LAy<rs Rnd Strvkes

Layer Layer Name Salient Services Provided by the layer

No.

-llandles physical and el ectrical Interfaces to the transmission medl um

2 Data nnk -Consists of two sublayers: logical link control (llC) and Media access control (MAC)

- llC: Formats the data togo on the medium; performs error control and flow control

-MAC: Controls data transfer to and from LAN; resolves conflicts with other data on LAN

3 Network Forms the·swltchlng/routll'lg layer ofthe network

4 Transport - Multlplexll'lg and de-multlplexll'lg of messages from applications

-Acts as a transparent layer to applications and thus isolates t hem from the transport system
layers

-Makes and breaks connections for connectlo,...oriented communications

-Data flow control In both directions

s Session -Establishes and dears sessions for applications, and thus minimizes loss of data during large data
exchal'lge

6 Presentation - Provides a set of standard protocols so that the display would be transparent to syntax of the
application

-Data encryption and decryption

7 Application - Provides applcatlo,...speclfic protocol s for each specific application and each specific transport
protocol system

OSI protocol architecture ITuly enables building systems with open interfuces so that networks using systems from
different vendors ore interoperable. Figure 1. 12 expands the bas·ic communication architecture shown in Figure
I .I 0 to an OS! model figure 1. 12(n) is a direct end-to-end communication model. The corresponding layers in the
two systems communicate with each other on a peer-to-peer protoool ioter.l3ce associated with those layers. ln
Figure 1.12(b), the end systems communicate with each other by going ·through an intermediate node/system.
Agnln. notice that the physical media connected to the end systems could be different. The intermediate s}'stem is
involved only up to the first three layers in the process. Layers 4-7 are not invo lved in the intennediate system.
This is analogous to a mail container with letters enclosed in envelopes being transported from one 10wn to another
town anywhere in the world. It does not matter what .network of intermediate cities (nodes) it goes through, or wbat
netwo(k of transportation media-surface, air, or water-it takes to get to the destination. The letter in the
envelope and contents of pa.ckages are untouched at the ITnnsfer points and are only handled by the sender and the
receiver; i.e., user applications.
Figurt I.U. OSI Communltullon Archlltc!lurr

EroS~Z

L
- ~-
-
-
--
,..._.,_

~-
_..._~fro~!_
... ~

$o......

T!a<UP!>'I
........,.
o-Ln< Datal lnlt

Pilylotol ~

I I

'*'"
I

--
~

r-
H-.rt<

DINII.iniL

The message in each layer is contained in message units called protoool data unit ( PDU). h consists of two parts-
protocol control information (PCI) and user da.ta (UD). PCI contains header information about the layer. UD
contains the data ihat the layer, acting as a service provider, receives from or transmits 10 the upper layer/service
user layer. The PDU communication model between two systems A and Z. including Lhe users at the top and IJ1e
u:ansmissioo medium at the bottom of the PDU layers, is shown io Figure 1.13. As you can see, the size of the
PDU increases as il goes tow81ds lower layers. If the size of the PDU exceeds the maximum size of any layer
specificauons, it. is the.n .fragmeote.d into multiple packets. Thus, a s ingle application layer PDU could multiply into
several physical PDUs.

Figu•·e 1. 13. PDU Communication Model between £ nd Sy.s1ems

I lh«A
I u-z I
l t
....
Ai>!~ic:at Applicotlon

-·- .
Prqwm1~ton Pleson"dtion

- L. d-x==z:3:
_..,
'Tf1tniiW"1

Nttwof~
-
-~·
fr~tnftJOI1
-..........
-
Dtlln Uno O<tto L~o~
-f>tol"lddi
(D}POU Oo1a SL~ltil~ ""~
~
~

l.5.2.1> rotocol La ye t'S :1n c.l Set·vices

We will OO\v go into some detail re~ding services provided by the seven layers of OSI protocols.

Layer I, physical layer, is responsible fur physically p.l acing the electrical sigonl on the physical medium and
picking up the signal from it. It controls and manages the physical and electrical interfu.c e.s to the physical medium
including the connector or the transceiver. The physical medium could be copper in the form of a twisted pair or
coaxial cable, optical fiber, or wireless media such as radio. microwave, or infi'ared. The signal could be either
analog or digital. There are various protocol standards fur a physical-layer interface depending on the transmission
medium and the type of signal. The 1wo classes of standards have been established by ITU-T and Electronics
Industries Association (EIA).

Layer 2 is the data link control layer, or data link layer for s hort. Data communication between two DTEs is
comroUed and managed by this layer. Note tllat in contrast to a. byt<>oriented tronsmlssion across a computer bus,
the data communication is a serial-bit-<>riented stream. The data link layer needs to do basic functions: first
establish and cle.ar the link, and second tra.nsmit the datn. Be.sides these, it also does error control and data
compression. Flow contro l on data link layer is done on a bop-to-hop basis.

For point-to-point communication using a dedicated facility, like the loop link from a customer telephone to the
telephone compa.ny switching office, the data link control is simple and stril.iglltfurwa.rd to implement. However, if
the DTE is connected to a LAN, or which is shared tf811smission media and is accessed simultaneously by many
users, then the data link control beoomes more complex. In the case of point-to-multipoint tmnsmission, the he!KI
end controls the access oft.he medium. LAN is a distributed e nvironment and tllus access control is di.stributed. ln
an OSI-layered model, the data link layer is divided into to,w sublayers-logical link control (LLC) and media
access control (MAC), as shown in Figure 1. 14. The lower MAC layer controls the access and transmittal of data to
U1e pllysicallayer in an algorithmic manner. There are three basic types ofLANs. Ethernet LAN is n bus type a.nd
the media is accessed using a distributed probabilistic algorillun. Carrier Sensing Multiple Access with Collision
Detection (CSMAJCD). The second type of LAN is a ring type U1lld in token ring (fR) and Fiber Distributed Data
lnterface (FOOl). A deterministic token-passing algorithm is used in this case. The third type. ofLAN is deployed
In wireless medium and is referred to as wireless LAN or WLAN. The probabilistic algoritlun, Carrier Sensing
Multiple Access with Collision Avoidance (CSM.A/CA), is used to access the medium . Random-access protocol
will be covered in Chapter 2.

Figure t .l4. Subtoytr Srruc:turt of • OAt.• Link l'rotoc:ol la~'fr

 Netwcrk

Logical Link Control

(LLC)

l\led1um Ace<:>$ Control

(MAC)

F'llysloal

LLC performs link management and dara transfer. Link management includes formaning the data ro go on the
medium, pe.rrorming error contro~ and flow control. If there is security required, it could be included in the ILC
sub layer.

The network layer is the third layer in the OSI protocol stack. It controls and manages the switching fabric of tb.e
·network. It provides both connectionless network service (CLNS) and oonnection-oriented ·network service
(CONS). The former is used when lower layers are highly reliable, such as LANs and bridges, as weU as when
messages are short. CONS is the method fur transmitting long messages, such as file transfer. It is also used when
the transmission medium is not reliable. It subdivide.s the transpol1 PDUs into !Tames of appropriate size based on
transmission parameters. The destinalbn address of each packet is read in both CLNS and CONS at the network
layer and routed 011 the appropriate link.

A router, or a routing bridge, at the nodes of a network perforras the function of routing and switching data. Any
subnetwork oftl~e node is under the control of that router. The subnetwork(s) can be anything &om a simple-single
segment LAN to complex subnetworks operating under a proprietary protocol. OSI architectural model handles
this by dividing the network layer into three sublayers as shown in Figure 1.15. The lop sublayer is the
Subnetwork-Independent Convergence Protocol (SNICP) layer Ibm interfaces to the transport layer. l 'he Internet
communicares between nodes using Internet address and SNICP. llte nodes in tum communicate with subnetworks
using the Subnetwork-Dependent Convergence Protocol (SNDCP), which depends on the subnetwork protocol and
could be any proprietary protocol In such a situation, the SNDCP communicates with its data link layer via the
third network sublayer, the Subnetwork-Dependent. Access Protocol (SND AP). This subnetwork arc.hitecture
isolates transpon and the above layers from the subnetwork dependencies. It also enables communication between
a DTE on dte lntemet and a DTE on a subnetwork node, as shown in Figure 1.16. Figure I. I 6(a) depicts network
configuratbn in which DT£..A connected to end node A communicates with DTE-N I connected to subnetwork
Mde·Nl via ·the intermediate system gateway node N. Figure 1.1 6(b) describes the path of communication through
different protocol layers from the originating end system to the terminating end system vta the intermediate node
gateway. The formats of the PDUs are identical in all three systems at SNICP layer levels and above. Access
networks having their own addressing scheme using Network Address Translator (NAl) or Dynamic Host
Configuration protocol (DHCP) can be implemented using tltis scheme.

flgurt l.t5. Sublayu Struclurt or a i'lrtwork Protoool Laytr

 Transport

SNICP

Networli SNDCP

SNOAP

Oatallnk

SNICP: SubnetWOtt-tndepend9nt Con~ergence Protocol

SNOCP: Subnetwork-Dependent Convergence Protocol
SNOAP; Sulrletwork-Oependent Adapler Protocol

Flgu•·• 1.16. Caccwoy Communlralion co Privoct Subnecwork

---1@

A-lh! s!Mda-d Nll!wcirl<

N· U1-N2-N3Sotlne!wor\ UI\Oef NOde N

ll'iln~ r- Trilf11!0f'll'1

SNICP

SNOOP
I~
SNDCP
SNJCP

OP-SN
SNJC
- SNIC?

SNOCP-sN

SNO~P SNDN' SNOA1'-S« SNCAP·SN

tataLmlti

1>11)""""
1- Dlll8Lnl<

rn,.~<a~
DO!> L111k·SI'l

PhyS)cllloSN
- t.la.Q LWI't •~

J J 1 1
N9twuu; Meot.tm S l.l1l00lwc:JIX Meaiul:n
(b) Pl'o4acol Olmml.,loa....,
The most used network protocol is tbe Internet Prorocol (IP) and has been PQpularized by the Internet. It is part of
the Internet suite of the TCP!IP and is a CLNS protocol. In OST terminology. It is called ISO-IP or ISO CLNP. A
connection-oriented OSI protocol is X.25 PLP, Packet Layer Protocol.

A popular scheme of implementing private subnetwork is to establish a network. with a private IP address, sucb as
1O.x.y.z. In this insrance, the gateway node, known as NAT, converts the global IP address to the local proprietary
IP address. fur example, LAN Z in Figure 1.8.

The transport layer is the fourth layer ofthe OSl protocol. It multiplexes the OD provided by application layerS and
passes packets to the network layer. Its se.rvicc is independent ofthe network on which the packets are transmitted.
The transPQrt layer can ag~~in be connectionless or connection oriented and is implemented in hoth Internet and
OS! protocols. As mentioned earlier, TCP is a component of the IP suite nod is conoection oriented. The
connect·ionless t·ransport protoool in a TCP/IP suite is called Lhe UDP. Plow control is also implemented in
tmnsport layers and functions as data rate manager between application programs and the network layer. ISO has
five transport layer specifications, TPO to TP4. TP4 is analogous to TCP.

Layers 5-7 arc application layer protocols. £xcepl in the OS! Reference Model, the tltree application layers are not
clearly separuted and independent. Let us look at each layer as if they were independent., like in the OSJ mode~ to
understand their specific fimctions and services provided. An application process commuolcates with flllother
application process during a seso;ion. The session layer services establish communication at the beginning of the
session. monitor, sync.hronill!, and error correct the information exchanged during the session. and then release the·
logical link at the end of the session. H is very strongly nilatcd to the presentation layer, which is the medium of
prese.:ttation of the context of the mess.~ge to the user or application program. In that sense, the presentation layer is
a context-sensitive layer. ll can be interpreted as the common language and image that the users at hoth ends of the
system use and understand- shared semantics of the two end users. A common abstract syntax that is used for
semantics is Abstract Syntax Notlltion Number One (ASN.I). Although the primary function of the presentation
layer is the conve.rsion of syntax. data encryption and data oompressio n are also generally done in that layer.

The lop and the seventh protocol layer is the application layer. The application 'Process interfaces with the
application support processes tbat are provided by this layer. Uke the other two layers in the set of application
layers (session and presenration), it is strongly coupled with the rest of the application layers. In the OS! Reference
Model, one can separate these processes &om the presenration and session layers, but in other models there is no
clear distinct·ion of the functions. Figure 1. 17 presents a comparison of the models-OSI Reference Model and
Internet model.

Frgu o•r t.t7. Conoroartson qfOSI anollnltmtll'roloc:ot l,ayrr M<>oltl(

 OSI INTERNET

Appllca ~oo

Appllcati.,.,.Spa:lfiCl
Pro~ rttnt.on
Prorocals

$<>10$000

TtlltllJIUO

fron!li>Ort ConOOCI...,.. I Conoll':lian·

le!!il: UOP orit~tlld. TCP

SNICP

N~l""'r~

I~
No!WC!1( IP
SNOCP

SNOAP

Oatn ~Ink
Not S(lft(!~l>d
l't1)Sital

The Internet model does not specey the two lower layers although H· is obvious Lhat !hey use distributed LAN and
WAN configurations. The tmnsport and network layers fotm the suite of TCPIIP protocols that we mentioned
earlier. Application layers arc combined into application-specific protocols.

Figure 1.18 shows a comparison offuur common applicat·ion-specific protocols in OST and Internet models. There
are more OSl applicalioo-specifie protocols, which we will not discuss here. All application-specific protocol
services in OS! are sandwiched between the user and pre-sentation layers. In the lntemet mode~ drey are
sandwiched between the user and the transport layer. The boxes on the right-hand side of Figure 1.18 describe the
comparable scrvites offered in the two model'!. A user intedaces with a host as a remote terminal using Virtual
Te.rmloal (VT) in the OS! model and TELNET in the Internet model. File transfi:rs are accomplished using File
Transfer Access and Management (FTAM) in !he OSl model and File Transfer Protocol (FTP) in the Internet. The
most common used mail service function in tbe Internet i<; Simple Mail Transfer Protocol (SMfP). A similar
protocol io the OS! model is the Message-Oriented Te-xt lnte.rchange Standard (M.OTIS). Network management is
accomplished using the Common Management InfOrmation Protocol (CMlP) in the OSr model and the Simple
Network MmuigemenL Protocol (SNM.P) in the Lntemel. We \viii extensively discuss tbe details of SNMP in thi s
book. CMIP is briefly discussed in Appendix for completeness. However, it is imporUlnt to understand tbe overaii
picture of protocol layers and other application protocols to appreciate network management fimctions that are
accomplished using network management protocols.

flgur• 1.18. APt>Utailon-SI!OciRc: Protorols tn OSI and lnitn>tl Mod•IS

 ~ ~
m - _J
{~I
,.--1---:c'--,.,.--'--~
( '::'I

1.6. Network~, Syste ms, alitl SL•n•ices

We described a network comprising nodes and links in Section 1.2. The phys.ical embodiment of a network·can be
deftned as a system. Thus, the nodes nod links are components of a network system. Just as a network can be
subdivided into subnetworks. a system comprises subsystems. A system or subsystem is made up of network
elements. Network elements can be either active or passive. Thull, a router is an octive network element, whereas a
splitter or a combiner !bat divides or combines signal energy is a passive element. A link could also be an active or
a passive component. 1n the case of an active transmission link, it can be subdivided into active nodes and passive
transmission media.

Servioes are fun~'lions ihat users derive out of networks and systems. Neiworks and systems exist to provide
service to the users. Service providers provide telecommuoicatlon services to subscribers and customers using
networks and systems.

J .(i.l . .Broutl blind Networks. Systems. 1m d Sl'n'ices

A broadband communication system can be defined as one that provides broadband service to homes and
enterprises. The common interpretation of this definition in practice· varies in different countries as weU as among
various service providers. ln the· most comprehensive defmition of !be term, we will define broadband
communication system as one that provides voice, video, and data services over the same medium to customer
premises. Broadband service comprising audio, video, and dara is also known as multimedia service.

Audio service includes telephone., telephone conference, and radio broadcast. Although the end terminals could be
either analog or digital devices, inrormation is carried digitally in the context of broadband service. A system
providing this service is tmly a real-lime information system.

Video service includes broadcast television, interactive television, video-oiHlemand, and video conference
services. Video service could be either real-time or quasi (near) real-time service. Once again, the presentation
could be on eitber analog or digital terminals.

Data service includes numerous applications, whi.ch can be classified into three categories: store-and-fof\vard,
audio streaming, and video streaming. Some· examp.les of store-and-forward service are email. messaging, and
Web-based applications. Audio and video broadcast and streaming services mentioned above sucb as MPJ and
video-oiHlemand can in a sense be considered under this category. They are not sensitive to absolute delay time
'between the source and the destination, but are affected by delay variations or jitter.

Broadband services are provided using broadband nehvo(ks. There are nume.rous types of networks to choose from
depending on what segment and what type of service one needs. It is like ordering ice cream in an ice-cream
pnrlor---oone orcup, hard or soft; size smaJJ/mediwn/large. choice of flavor, choice of topping, etc.
The Lhree segmenLsofbroadband network are WAN, broadband access network, and CPE nemmk.lobroadband
tenninology, the CPE network is also called home network when the customer premises is a residence. Network
segments and choices in various segments are shown in Figure 1.19.

Flgur • 1.19. Broadband NtlworkS.gmtnb and Ttcbn ul ~gltll

INIIIII
-··
llloliol:><l<

The WAN and access network interface with each other via !he edge rouLer. The demarcation point between the
access network and CPE network is shown as the residential gateway. Although 1his is the logical demarcation
point the physical de.marcation point between the access network oft be service provider and the customer-owned
CPE, or home uetwork, could be different. As an example in the cable network, the demarcation point is called
Network illtcrfuce UnJL (NTU) or Network lnterfuce Device (NID) and is the physical termination of the cable
access network outside the house. Tbe residential gateway·may onnay not·exist, and if it does, it is a part ofCPE
network.

1.6.2. Wid e Area etwork<i

The four leadmg networks and protocols that are used in broadband WAN are Internet using Asynchronous
Transfer Mode (ATM), Synchronous Optical Network (SONB1), IP, and Multiprotoco.l Label Switching (MPLS)
network.

ATM. netwo rk: ATM network is ideally suited for WAN or core network. It has fast. layer 2 sw itches that can be
configured to func1ion in parallel and thus can process bigb data rate cell-oriented packets. l...alc:ncy can be set in
ATM. switches by setting priorities to tbe different services-real-time and non-real-tim~!Kling provided
Furthe-r. traffic ·perfbrmance is iooreased by establishing Virtual Path-Virtual Circuit (VP- VC).

Four classes oftruflic have beeo defined in ATM. network to implemenL quality of service. COostaol bit rate (CBR).
real-time variable bit rate (VBR-R.T), oon-.real-time variable bit rate, (VBR-NR.T), and available blt rate (ABR) or
user bit rate (UBR). Transmission of voice is assigned CBR. An example ofVBR-NRT is transmission of stiU
images. Data 1raffic and store-and-forward traffic get the lowes! priority, AB R.

SONEf: An optical fibe.r medium can be used to carry multiplexed lower bandwidth signals implementing SDH.
This mode of transmission is known asSONET. The Oplicaltransmission network oontains regenerators, digital
cross·oonnect elements, and add-and-drop multiplexers (ADM). Modem optical networks use dense wavelength
division multip.lexers (OWDM) and very high bandwidth signa Is can be transmitted. through dtis optical network.

Internet: The lntemet backbone WAN using IP is highly matured. bas a full set of application·o riented li:atures,
and can interface with access and CPE network fn a more seamless manner. However, its main drawba.ck is that it
is difficult to meet qualit.y-Qf-service requirements needed for multimedia broadband service. Because of its
variable packet size. and packets choosing possible alternate padts between the source and the dcstin!llion, th.e
performance of routers and other transmission devices is not as efficient as in an ATM network.

Quality of service in IP-ori.ented WAN traffic is improved by implementing one oftwo different approaches. They
arc integrated service [RFC 2205] and differeotlated serv.ice [RFC 2474]. Jn one form of implementation, lntserv
packets in dte·lntemct are classified into dtree classes: guaraateed, controUed or predictive, and best effort. Intserv
rese.rves bandwiddt from dte source to dte destination on a pe.r-flow basis fur a guaranteed class-of-service call or
session using reservation protoool, RSVP. Once the reserved path with dte necessary bandwiddt is established, dam
arc transmitted. The bandwidth is released after the calVsession is completed. lntse.rv is not an efficient scheme for
establishing quality of service in the backbone network as the.re is no guaran.tee that the resources will be available
wben needed. Furdter, the scheme does not scale weU.

In the diffi:rcntiated service, diffserv, packets belonging to the same class are grouped at each hop and then
prioritized. There are four classes and each class bas dtree subclasses for dropping packets-low, medium, and
bigb. The present tre.nd in providing quality of sc..Yvice ror backbone is to use diffe.reotiated service comple.mented
with some form of reservation capabilities of RSVP.

MPLS network: MPLS attempts to oombine the benefits of ATM quality of service with feature benefds of the IP-
based Internet. Conventional routers examine the packet headers and classify them into forwarding equivalence·
classes (FE C). They are then assigned the next hop. 1n MPLS this is done once; possibly at tbe ingress router, and a
label is attached to it. At each router, only the label lookup is done for detem1ining the next bop. Label lookup can
also be done using a switch. A router that silpports MPLS is known as a Label Switching Router (LSR). MPLS can
support nny network layer protoool. RFC 3031 describes MPLS nrchitecture fur an IP network layer protoool.

1.6.3. Broutlbund Access Networks

Figure 1.20 shows six types of broadband access networks dtatprovide broadbiiJld service to homes, Small Office
Home Office/Small and Medium Enlerprise (SOHO/SM£), and enterprises. The core network is ll'/ ATM/MPLS
WAN. The l.ink from the head e.nd or dte edge router to business customers is shown as an optical earrier-n (OC-n)
link, afthough it could be any other transport scheme. Hybrid fiber coax (HFC) cable network and Digital
Subscriber Line (DSL) nelwodt are dte matured access networks. Fbted wireless is bei.ng offured as point-to-
multipoint service or meshed oetwork. W!Max, 10 metropolimn areas. Mobile wireless could be offered using
either 3G technology or wi~less !.,AN. The furmer has the limit <Ilion on data rat.e and the latter on range. Fiber
network as Passive Optical Network (PON) is still in an embryonic stage fur economic reasons.

Figuc·• 1.20. BroiKlb•od A<tt5!l N<IWttrks

Cable Access Network has its head eod interfacing to t.he edge router. Analog and digital signals from various
service.s are multiplexed at the head end and are converted &om on electrical signal to optical wavelength signals.
The optical signal is then carried over fiber up to an intermediate point, optical node, where it is dow~Konverted to
radi:> frequency and transmitted the rest of the way to the cust"omer premises over two-wny coaxial cable, hence the
term hybrid fiber coax (HFC). At the customer premises, the TV analog signal is split from the digital data. The
latter is demodulated to a baseband digiml signal using a cable modem and is fed to the digiml devices, such as
computer and appliances.

Digital Subscriber Une access ·network uses a telephone line and can be deployed uSing different
implemcnmtioos, refe.rred to as XDSL. Of tl~ese, Asymmetric DSL (ADSL) shown in Pigure 1.20 is the most
prevalent deployed all over the world. AU hough cable network is more commonly used in the United Smtes by a
ratio of approx.imately 2 'to I, the reverse is the case in the rest of the world. The technology uses the .existing
unshielded twisted-pair (UTP) wire that carries the analog voice to transmit data in addition to voice. The voice is
carried as an analog signal at the low end of the frequency spectrum (0-4 ld:lz) and the digital dam over the higher
band of 1he spectrum. It is termed asymmetric as 1he downstream data rate (from the centra I office to customer
premises) is much higher than the upstream (1iom customer premises 10 the eent.ral office) data rare. The analog
voice and digital dam are separated at both e.nds of the aocess network using a fiker, and the digital dam are
modulated and demodulated at both ends using ADSL modems. Atlhe central office, voice circuit interfaces with
U1e central office ~·witch and the digit.al daia with the edge router.

Wireless Access Networks: Figure 1.20 shows three types of wireless access networks. The terrestrial wireless
network, also known as fixed wireless, is a point-to-multipoint transmission. A base smtion with mullipleantellll8S
covers multiple sectors, each serving many subscribers. The two well-known deployed technologies are
Multichannel Muh:ipoint Distribution Service (MMDS) for rural areas and WiMax fur urban areas. Satellite
wireless syste.ms are primarily used for on~>,vny televisi:m broadcasting service. Mobile wireless has limited
bandwidth and is currently used In phones such as smart pbooes, providing bl:oadband service.

1.6.4. Homf/CJ>E etwo rks

CPE network in enlerprise environmem is eitber an IEEE 802.3-based Etbernel LAN or I:EEE 802.11 based
wireless LAN, also known as W!Fi, or a hybrid of both. Horne network provides the opportunity to uti.l ize mulliple
technologies besides Etbernet LAN and WiFi. HomePN A is implemented using twisted-pair telephone cable
medium, HomePiug takes advantage of power line wiring in the house, and cable utilizes ahe aelevision coaxial
cable. Fire Wire is also a wired medium and is based on IEEE 1394 protoco l to trans mit high-speed digital dala
Universal Se.rial Bus (USB) is u$00 for low dalll rilte peripherals. Wireless home network technologies include
Blueaooth and ultra-wide band (UWB) personal area networks (PANs) fur short distances.

1.6.5. Quality of Sc!"icC in Broadb"Jul Syst ems

Q uality of service CQuld be interpreted in toohnical terms in many different ways. However, from the users' point
of view, people are used to reliable, dependable, and good quality analog telephone and tcle,~sion sc.r vicc. They
expect the same quality of service when the telecommunication and cable services are extended to broadband
service that inchJdes voice, video, and data. Networking aeclmoiogy has to prioritize real-aime voice and video
traffic over store-and-forward data traffic, and provide the end-to-end quality of service. For real-time applications
of voice and video. the delay and jitter should be imperceptible. Service should be highly dependable (always
available) and reliable (quality is consistent). Monitoring and manag.ing these parameters is achllllenge for network
management.

1.6.6. Security und Privacy in Broud bund Systems

With universal 10 and multiple service providers delivering multiple services on shared media to multiple
subsCribers, the security and privaey of information becomes a primary concern. This is especially critical with e-
bu~iness over the Internet. Besides implementing security and privacy-authentication, authorization, and
encryption-of tbe· data and management infunnation, tbere has to be a cultural change in the perception of the
subsCribe.r s that the information link is secure.

1.7. C nse Hisfo-rics on Netwo rk, System, and Service M:1nagc ment
Network Management is more than just managing the network. In standards bodies it i.s referred to as Operations,
Administration, Maintenance, and Provision ing (OAMP). Of course, networking and network management existed
befOre network management became a formalized diScipline. Network management and its complemcnlllry
functions ofsystem ~management and application management are all means to tbe end ofservice ~management in
provi:ling the subscriber or customer quality of servi.c e. As one IT manager commented, the configuration and use
of a NMS formalizes what a network administrator would have otherwise. done. llte network administration "war
stories'' in the folJo,ving subsections illustrate thai network management (especially without proper too ls) could
presenl a challenge to IT managers.

J.7.J. Case History 1: lJUr)Ort:wce of ToJ}Ology ("Case o ft be Footprint}

A saable corporate n~'lwork consisting of several minicomputers arJd about 100 desktop wo rkstations and personal
computers suddenly started "crashing" freque.ntly (a legacy network·e x.ample). How often have we heard 11 network
coming down without any apparent reason? Here is bow one· Vice President of l.nrormation Systems desc.ribes an
incident.

Part ofthe network went down in the engineering area one morning. Since there were a whole series ofusers and at
that time we were not using a STAR ( hub) topology. but ratber the old-fashioned serial topology (wbere all the
users were daisy cbnined to the coax), we suspected a break in abe cbain, probably at a tr.ansceivcr lap. Lacking
sophisticated NMS tools, Infurmatk>n Systems persorutel started walking dte hallways asking the users If anyone
bad just been doing anything o ul of the ordinary, which might have broken tbe chain and caused the problem.

The guys came back and reponed tbat no one hnd said abat they bad ''done anything." So I (VP) started back down
the halls with the guys and peeked into each office. Finally, I stopped and said "Let's look up in the ceiling bere."
Sure enou,gll, we found a transceiver that someone bad been fooling with and tbat was not properly connected,
which bad caused the break. Once connected, the network segment came bac.k up.

The guys asked "Why did you say- try here?" particularly since the engineer in tbat office claimed ignorance. I
calmly pointed to a dusty image of a sneaker footprint on the engineer's desk and the ceiling tile that was ajar
above the desk and said-"you need to use all the diagnostic tools at your disposal!"

1.7.2. Case Hi~tory 2: Centrally Managed Network lssuc.s

There are numerous war stories tbat we can describe relating to heavy load on a NMS managing the network and
network e.lements. We will choose one tbat illustmtes several is.sues related to network design, configuration, and
maintenance. An integrated network management system (INMS) was integrating alarms from multiple element
managemenl systems (EMSs) in a service provider network. Each EMS manages a domain of network elemenlS
and passes the relevanl events to the INMS as sbown in Figure 1.21. The service provider is able to monitor in its
centrally located NOC faults occurring in its global network. As simple as this sounds, its imple.ment;ltion could be
extremely complex. Let us consider a si,mple real-world s ituatioo in which a f'ew EMSs were integrated into an
INMS and the alarm occurrence time in the INMS was at variance with the individual EMSs.

Flgur t 1.2t. cas·t History 2: C<nlr olly Maoagtd Nthmrk lssu ..

EMS EMS

Each EMS records and displays tbe receipt time of tbe alarm. The same is transmitted lo tbe INMS. lt was
observed tbat tbe indication ofibe time at wbicb the alarm occurred was significantly different in INMS from that
indicated in the EMSs thru were sending the alarms. The alarm occurrence time was ooosidembly delayed,
sometimes by hours, in INMS. Tite cjJaUenge in a centmlly managed .network is to find the root cause of the
problem. Is it network delay? Is the delay due to excessive number of events? Is it due to input/outpur (1/0)
limitation of the input port of the lNMS? ls it due to I/O output port of EMS? ls it in the software of either EMS or
INMS or both? lH is in the INMS software, sbould the filtering of unnecessary events at the input take care of the
problem? The answers io most of these questions were affirmative for each. but to a varying degree In each case.
The predominant cause is the stress on NMSs, although it can be traced sometimes to network elements in the
various domains. Tmnsmis.sion of unnecessary alarms also causes a stress on tbe network and networks have gone
down due to uncontrolled generations.of network management messages.
I.7.3. T r ansaction Delays in Client-Ser ve r Netwo rk

In cumru oational and global enterprise organizations, application servers serve thousands of clierus over
international nelwork.~. In a study of banking industry, lrnnsaction del11ys were measured and analyzed to determine
the root cause ofthe de illy as reported by tellers of bmnches. The propagation time of individual trnnsaclions was
monitored as they traversed through the LAN networks and servers of the branches, through the WAN, and
centra.lly processed by an application server. Some of the transactions were discovered to time> out due to long
tra.nsaction delays. Study results identified the source of lhe problem to be galeways and applications; and
appro priate actions were initiated to resolve the problem. This case illustrates the need tor management of end-to-
end communication and the influence of network components, applications, and client-5erver arohitecture in a
network.

1.7.4. Service lmp:lct in End-to-End ScJVicc of CustOmers

End-t<rend communication is further illustrated by the need to proactively identify the service of the customers
affected by a network element failure·. This is illustrated by the following case. In an optical fiber transport network
using TDM SOH network element that carries thousands of cbnnnels, Ihe failure of a single component affec1S
services of hundreds of customers. An end-to-end communication breakdown is to be lraced to the failure of a
single or multiple network elements by root cause anliJysis and dyii1U11ically determine all clients whose services
are impacted. The serv.ice provider detects the problem even be!Ore cusfomer complaints nre received and in!Orms
the customers that the problem is already being addressed to restore service as soon as poss.ible.

l.7.5. Some Common 'etwori< l'roblcms

The most common and ser_ious problems in .network ore connectivity failures and are .bandled under tbe category of
fault management. Fauk is genemlly interpreted as fuilures in accessing networks and systems by users, Network
failure is caused more often by a node failure than failure of passive links (except when it Is cut by construction
crew). Even node failures are more often limited to specific interface failures, When thi s happens. all downstream
systems from that interface are ioaccessible. Such tailures are associated with fuilure of the network interface card.

Node failures manifest as connectivity failures to the user. There are networking tools ava.ilable to the manager to
localize the fiml~ as we shall learn in Chapter 9 on Network .Management Systems and Tools.

Another cause of nerwork conoect:ivity failure is procedum~ bul very common. Network connectivity Is based on
the IP address, which is a logical. address assigned by the network administrator. The lP address is uniquely
associated with a physical MAC address of the network component. However, mistakes are made in assigning
duplicate!P addresses, especially in an enterprise environment with mukiple system administrators,

A host or system interface prob lem in a shared medium can bring the entire segment down, somet imes
intermittently, as shown in Case History 1 abo\<e. 1l1is could be a nightmare for the network manager to isolate
without causing lnlerruption in service. A network roaoager uses intuitive knowledge to look for patterns such as
change in configura.! ion, addition of new equipment or facility, etc. in resolving such problems.

Intermittent problems could a.lso occur due 10 traffic overload causing packet loss. Sometimes the management
system may indicate fill lures, when in actuality data traffic is nowing normally. Perforroance mooitoring tool.s
coukl be useful in tra.cking such problems,

Power hits coukl reset network oomponent configuration, causing network ful.lure. The network has a permnnenl
configuration (default) and. a dynamic configuration (run-time). and. thus a power hit could chaoge the
configuration.

Finally, there is the non-problem, which really means that the cause of failure is a mystery. Tbere is nothing else
that a network manager could do except tum the sys!em off and then on. Bingo! The problem is resolved.
Perfonnance problem could also manifest as network delay and is more an annoyance to the network manager,
who needs to separate network delay from the application program or application processes de.lay. Then the
network manager ha~ to convinoe the user and then the person responsible fur the application to rectify the
situation.

With the ever-increasing size of the network and connectivity to the Internet, security violation in network
management is a frequently encountered. problem. This is more a policy problem than technical, which we will
address in Chapter ll when we discuss security management.

1.8. Challenges ofiT Managers

Managing a corporate network is becoming lutrder as it becomes larger and more complex. When we talk about
network management, il includes not only componeniS that transport information in Lbe network. bur also systems
that generate rraffie in tbe network. What use is a computer network if there are no systems in dte network to
provide service to users? The systems could be hosts, database servers, file servers. or mail servers. In the client-
server environment, network control is no longer centralized, but distributed. Computer and tek!communication
networks are merging fast Into converged nct\vork with common modes and media of transportation and
distribution. As in the ca;;e of broadband networks, the IT-manager needs to maintain both types of networks. Thus,
the data communications nutnager func.tions and telecommunication manager functions have been merged to I bat
of tbe IT manager. With t:be eKplosion of information storage and transfer in the modem information era,
management of information is also the responsibility oftlte· IT manager, with the title ofCIO, Chieflnformation
Officer. For example. the IT manager needs to worry in delall about who can access dte infurmation and what
inlbrmation they can access, i.e., authentication and authorization issues of security management. The corporate
network needs to be secured fi.)r privacy and content. using firewalls and encryption. Techno logy is moving so fast
and corporate growth is so enormous, that n CIO has to keep up with new technologies and the responsibility lilr
financial investment thal the corporation commitS to. This amount.s to millions ofdollars, and the success or failure
of making the right guess-not choic-.:ould make or break the CIO's job. Notice that dte word "guess'' was use.d
instead of "choice" deliberately because it is not always clear wb.iob oftbe options nre a dead end, a.nd hence need
to be avoided. Since they nre not obvious, the IT manager needs to make provisions for contingencies to change
direction when the IT industry does.

A good example of indeterminacy in the fast-moving tecbnology industry was competition between the two
technologies of Ethernet and ATM to desktop. ATM was pr<XIictcd to be tbe way to go a few years ago. However,
·this has not been the case because of tbe development of enhanced capability and speed of Ethernet. Anotber
current example related to this is tbe decision that one has to make in the adoption and deployment of WAN-
whether itshouk! belP, ATM, or MPLS.

Perspectives of Network Managers ln order to appreciate challenges t:bntlT nutnagers face, several of them were
interviewed by the author. They face net\VO(k administration and manageme.ot problems day in and day out. These
are the folks who carry a cell phone with them all the time since most corporate networks run 24/7-i.e,, available
24 hours 11 day 7 days a. week! The questions Uta! were posed, wiU1 n summary of t:be answers edited ror the current
SLatus of IT, follow. They are not an ellhaustive list of queSI·ions and answers, since thai would make the contents
of a separate book, but are<mly intended tc;> indicate the compleKity of managing a network and thus motivate a
student in networking. Notice that it is not just a technical function, as Case History I exemplifies. Also, e\<en u.se
of t:be beSI NMS does not solve the problems associated with building and maintaining a network, but it is a
necessary tool. Thus. learning network management involves more than understanding net\vork and network
management protocols. The author'·s recem.in-deptb study of service providers alsp raises similar comments.

General

People el<pect a network to function like a telephone network.

 Reliability in a data network as in a telephone is unrealizable. The telephone network was monopolistic and
had expensive redundancy. llte data network is ad hoc, decenu:ali7..ed, has loosely specified interfaces, an.d
has dynamic routing. Thus, it is a lot more flexible than the telephone network though less reliable.
• Designing, deploying, and managing net~vorks that can handle reol-'iime and non-real-time dalll.
Integration ofmultivendor 811d multitecboologyequipment and their network management systems.

I. Wbat are your top challenging activities in managing the network?

o Rapid advance oftechnology
o Problem analysis-needs human intuition and skill besides scpbisticated management too ls
o Anticipate customer demands
o Acquire and retain human resources
o Manage client~rver environment in converged networks
o Networking with eme.r ging technology necessitates the need for continuing education
o Collaborative research between academic Institutions and industry
o Maintain reliability, that is, make changes, upgrades, etc. without disrupting the network and
impacting business
o Diagnose problems or outages in a non-disruptive maaner (without impacting other users on the
network)
o Bstimate the value of a technology transition. For example, should one transition over to
accommodate the increasing number of!P addresses with !Pv6 o r oontinue with !Pv4 with Network
Address Translation (NAT) as a hierarchical addressing scheme?
2. Which elements of managing your network require most of your time? What percentage of time do you
spend on maintenance compared to growth?
o A 30-SO% growth, 20-70% maintenance b~o;ed on the organization.
o Configuring the management system itselflllkes most of the time.
o Expanding the network.
o Gathering and analyzing statistics for upper management review to conduct business.
3. How did you or would y.ou manage your network without an NMS?
o Reactively, .not proactively; firefighling
o Troubleshooting tools, e.g., sniffer, ping, etc.
o Home-grown systems using an open scurce, e.g., Multi Router Traffic Grapher (MRTG)
o Rely on consultant advice and technical informal ion for growtb decisions
4. Do you need an NMS? Wby?
o Por proactive management of network
o Verify customer configuration
o Diagnose problems
o Provide statistics on performance
o Help remove bottlenecks
o NMS formalizes the manual practice of network management
o NMS products reflect the company's practice that develops them
o To see theirend in growth
5. What problems would you expect the NMS to resolve, and how?
o Bnhance customer satisfuction by meeting the Service Level Agreement (SLA)
o Save time and poople rescurce and thus cnbance productivity
o Turn-around shoner fur re,sclotioo of problems
o Gather statistics and predict trends for planning purposes
o Do curnent events
o Troubleshooting
o Remove constmints and bottlenecks
o FouJt isolation
o Expect the NMS ro do a root cause analysis and pinpoim failures

We will now briefly introduce the subject of network management funetionsand system in the following ;ections.
1.9. Network Ma nagement : Goals, Orga.nization, and Functions

Network Managemenl. can be defined as Operations, Administration. Maintenance. and Provisioning (OAMP) of
network and services. The Operations group is concerned with daily operations in providing network services. The
network Administration is concerned with establishing and administering overall goals, policies, and procedures of
network management. The Installation and Maintenance (l&M) group handles functions that include both
installation and repairs of lilcilities and equipment Provisioning involves network planning and circuit
provisioning, traditionally handled by the Engineering or Provisioning department. We will describe each of these
functions in this section. Although we continue to use the tenninology of network management. in r.he modern
enterprise environment this addresses all om and IT services.

1.9 .1. Goal of Networ k Management

The goal of network management is to ensure Lhattbe user.> of network are provided IT services with a quality of
service that they expect. Toward meeting this goal, the management should establish a policy to either formally or
infOrmally contract an SLA with users.

From a business administration point of view, neh,wk management involves strategic and tactical planning of
engineering. operations, and maintenance of netv.'Ork and network services fur current and future needs at
mirtimum overa U cost. There needs to be a well-established interaction between the various groups performing
these funct ioos.

Figure 1.22 presents a to]Hiowo view ofnetwork management functions. lt comprises tltree major groups: (i)
network and service provisioning, (ii) network and service operations, and (iii) network r&M. It is worth
considering the different functions as belonging Ill specific administrative groups. alr.hough there arc other ways of
assigning responsibiliiies based on loenl organizational structure. Net~vork provisioning is the primary
responsibility of the Engineering group. The Customer Relaiions group deals with clients and subscribe.rs in
providing services planned and designed by the Engineering group. Network l&M is the primary responsibility of
r.he Plant Facilities group. lnter:-dCtioos between the groups are shown in Figure 1.23. Nonnal daily operations are
the function oftbe Network Operations group, which controls and administers a NOC. This is the nerve center of
network management operations. The functions of NOC are primarily concerned with network operations; its
secondary responsibilities are network provisioning and network l&M. The associated service operations are
handled by a subscriber operation ceruer (SOC) and customer relations management (CRM). Our focus here is on
NOC.

Figurt L22. Nrtwork Manugrm<ul Functionlli Groupin ~

Flgul'f 1.2.l Nttwork Man-agement Fw1etional Flow Char1

1.9.2. Network J>rovisioning

Network Provisioning consis1S of network planning and design and is the responsibility of the Engineering group.
The Engineering group keeps track of new technologies and introduces them as needed. What is needed and when
it is needed are determined from analysis of traffic and pedilrrnance data provided by the netwo.ck operations. New
or modifications to network provi sioning may also be initiated by management decisions. Planning and efficienl
use ofequipment can be acbieved w.itb good inventory management of current and fitture modification~ of network
configuration by tbe Network Provisioning group.
Network management roots are helpful to the Engineering group in gathering statistics and studying 1reods in
traffic patterns fur planning purposes. Automated operat·ions systems help in the design of circuits and measuring
the performance tune-up.

1.9.3. Network O twmtintli and 'OC

The functions of network operations listed in Figure 1.2'2 are administered by the NOC. They are concerned with
daily opemtions of the network and ·providing ·network servi.ces. lSO has defmed five OSI network management
applications, which are fault, configuration, performance, security, and account management. They are also
responsible for gathering statistics and ,generating reports for management, system support, and users. NMS and
tools are a necessity for NOC operations. They are U$ld in various management applications described below.

Fauh Manageme.nt/Service Restomtion: Whenever there is a service failure, it is NOC' s responsibility to restore
service as soon as possible. This involves detection and isolation of the problem caus.ing the failure. and restoration
of service. In sever:al failure situations, the networlnvill do this automatically. This network feature is called self-
healing. In other situations, NMS C41n detect failure of components and indica.te with appropriate alarms.
Restoration of service does not include fL'Iing the cause oflbe problem. That responsibility usually rests with the
I&M group. A trouble ticket is generated and fOllowed up for resolution of the problem by the I&M group.

Trouble Ticket Administration: Trouble ticket administration is the administrative part off.~uh management and is
used to track problems in the network. All problems, including non-problems, are to be tracked until resolved.
Periodic. analysis of the datn, which are maintained in a database, is done to eslllbli.sh patterns of'the problems for
fOllow-up acl'ion. There are troubl(.}otracldng systems to automate the tracking of troubles from the automatic
,generation of a trouble ticket by an NMS to the resolution of the problem.

Confrguration Mnnagement: There are three sets of configuration of the network. One is the static contiguratbn
and is the permanent configuration of the network. However, it is likely that the current running configuration,
which is the second, coukl be different from that of the pem1811ent conftguration. Static configuration is one that
tl1e network would bring up if it is started from an idle status. Tbe third corlfiguration is the planned configuration
of the future when the conftguration data wiU change as the network is changed. This information is useful for
planning and inventory management. The configuration data are automatically gathered as much as possible and
are stored by NMSs. NOC has a display that reflects the dynamic oonfigurat bn of the network. and its stows.

The status of the network is displayed by a NMS and indicates any fitilure of components of the network, as well as
·the traffic pattern and performance. Any configuration changes needed to relieve temporary congestiln in t ra ffic
are Olllde by NOC and are reflected in the dynamic display at NOC.

Performance Management: Data need to be gathered by NOC and kept updated in a timely fashion in order to
perform some of the above functilns, as well as tune the network. for optimum performance. This is part of
performance· management. Network SUllistics include data on traffic. network availability, aod network delay.
Traffic dam can be captured based on volume of traffic in various segments of the network. They can also be
obtained based on different applications such as Web traffic. email, and network news, or based an iriUlSport
protocols at various layers such as TCP. UDP. IP, rPX, Ethernet., TR, FDD[. etc. Traffic statistlcs are helpful in
detecting trends and planning future needs. Performance data on availability and delay are useful for tuning the
network to increase the re Uabil ity and to improve its response time.

Security Management can cover a very broad range of security. J't involves physically securing the network, as well
as access to the network by users. Access privilege to applicntbn software is not the responsibi lity ofNOC unless
·the application is.either owned or maintained by NOC. A security database is .established and maintained by NOC
for access to the· network and network information. There are other aspects of security management such as
firewall~ and cryptography, which \viii be introduced later in Chapter II .
Accounting Management administers cost allocation of the usage of network. Metrics are established to measure
the usage of resources and services provided.

Since the network consists of compo.nents manufactured by muliiple vendors, commonality in the definition and
relationship of component attributes is needed. This is defined by Management Information Base (MIB), which we
will discuss in Pan II. Some of the data acquisition has to be manual (because of leg~~cy systems), but most data
can and should be acquired in an automated mode. The SNMP is the most popular protocol to acquire· data
automatically using protocol- and perfurmance-analyl.ing tools.

As pan of implementing the above standards, we need to ensure that adequate reports are generated and distributed
to relevant personnel. There are, in general, three classes of reports: SYS!ems, management, and user. SySlem
reports are needed for network operations to track activities. Management reports go to the managers of network
management group to kee p them infunned abo nl the activilies and performance of NOC and the network. User
reports are distributed to users an a periodic basis o r are available on-line to let ihem know the stains of network
performance.

J.9.4. etwori< ln.stallation 11.utl Maiuteuaoce

The Network l&M group takes care of all activities oflll$tallation and mainie08Jice of equipment and tnUJsmission
facilffies. 1ltis group is the service ann of the Engineering group for installation and fiXing troubles for network
operations. The group works closely with the Help Desk in responding to the problems reported from the field.

Having introduced what network management is from an operations, administration, maintenance, and planning
viewpoint. let us next oonsider the architecture and orgait ization ofan NMS.

1.10. Nclwork 1\hnagcmcnl Archilcclu rc and Org11ni7.a1ion

We need to distinguish at the oUlset the difference between network management and network system and service
management. Remember tbat a user may not mnke that distinction when be or she cannot access an application on
a server from a client application in his or her workstation. This could beeither due to a problem in the application
progrnm in the server affecting one o r mare clients or due to a.lrnnsport pro blem.from the client workstation to the
server platfonn. The fanner is a network system problem affecting tbe serviceofrered and ralls under the category
of network system and service management. T he latter is a connectivity problem and falls under network
management. We can genernlicre system and service managemeni as the mnnab>emem of systems and system
resources in the network and services ofrered by the network. Network management is concerned with network
resources such as hubs, switches, bridges, routers, and gateways, and the connectivity among them via a network.
l t al so addresses end-to-end connectivity between any two processors (not application processes) in the network.

As we saw in Section 1.1, a network consists ofnet\'"r.k components and their interconnection. Each vendor, who
manufactures a n etwork component or a set of network components, is best qualified to develop an NMS to
manage that product o r set of products. T his invo lves getting data from each instance of tbat. compo nent in the
network to one o r more ·centralized locat ions and displayi ng their status on an NMS; for example, failure of a
bridge . This would set up an alarm in the NMS to alert operat ions personnel of the fai lure. This would enable
operations personnel to follow up on the problem and restore service·, even before the user calls in a complaint.

As mentioned above, e!ICh type of component is managed most efficiently by its respective management system.
There is need for an NMS to manage all the components tbat are connected to a net\vork. Ag~~in, it is relatively
simple for a vendor to develop an NMS to manage a network comprising only their components. However, a user,
sucb as a g.lobal corporation. buys components from many different vendors, and the lnfonnalion systems manager
oftbe corporation bas the responsibility ofma inta.ining the network of all vendor components. 1ltis mi~ht require
the installation of multiple NMSs fur an e nterprise o r an NMS tbat can manage multiple vendor components of a
network. Thus, common management system, as well as the integration of different management systen\s and the
interoperabilffy between them, has played a major role in the network management arena. Standards org~~niz:ations
and industrial commun.ities bave est'l!btished standards for this purpose, whiob are still eva lving. The two major
management staodards are the lnternet developed by the lntemet Engineering Task Force· (IETF) and OS!
developed by the ISO. We will look at the former in detail in this book. There are also standard~ that are developed
by industrial consortiums associated with specific technologies, such as DSL Forum and Cablel..abs.

Network management dumbbell arehitecture for interoperability is shown in Figure 1.24{a) where two vendor
systems A and B exchange common management messages. The messages consist of management infurmation
data (type, id, and status of managed objects, etc.) and management controls (setting and changing configuration of
an object). The protocols and services associated with dumb beU architecture are presented in Figure 1.24(b).
Application servioes are the management-related applications Such as fault and configurailin management.
Managemem protocols are CMJP fur the OSl model and SNMP fur the Internet model Transport protocols ore the
first four OS! layers for the OS! model and TCP/lP over any of the frst two layers fur the l.nternet model

Figw·• t.24. Nttwork ~bnagtmtnt Dumbbtll Architrcturt

8 8

Figure 1.25 models a hierarehical configuration of two net\mrk agents monitoring two sets of managed objects.
The agent could be an embedded agent in a network element or nn EMS communicating with agents embedded in
the network el.emeots. Ao NMS is at the top of the hierarchy. Each network agent monitors its respective objects.
Bither in response to a polled query from the NMS or triggered by a local alarm, the agent communicates to the
NMS the relevant data.

Frgurt 1.25. Nttwnrk Manogtlnrnt Comt>Onrn ll!

Peer networks can communicate network management messages and controls betw~ each other, as shown in
Figure 1.26. An example where such a configuration could be implemented would be two NMSs associated with
two telecommunication networks belonging to two network servk:e providers; for example, an lnterexobange
cll!T.er and a local access provider. As the i wo NMSs communicate with each other, each NMS can superimpose
the data from the other and present an integrated pit1ure to the network administrator.

Figure 1.16. Ntlwork Mau•g<mtnl lultrot>•nlblllly

We want(o make one final note before we leave Utis section. Some of the issues associated with the management
of telecommunication network by the telecommunication service providers are unique and involve more than just
management of networks. lltis has given birth to the Telecommunication Management Network (TMN)
framework and related slllndnrds. We will address these in Chapter I 0.

1.11. Network Ma nagement Perspectives

As we said earlier, the NMS primarily manages the networks that transport infoonation. However, from a user's
perspective, networks are means to an end, namely to have access to infonnation ucross the networks. Thus, the
users' needs require a total solution to mana&re the networks, system re.sources, and appucations that nm on
systems. Applications coukl be specific user applications, or general-purpose servers such as file servers, database
servers, and DNSs. Software products have since· been developed to address such system- wide so lotions.
An lT manager is interested in more than managing networks, sys1ems, and applications. l:!e or she would like to
amoma1e other functions such as back up of databases and programs, downloading of software updales from a
centml location, and a host of other support functions. These are required to run an IT opemtion efficiently and in a
cost-effective manne r.

Another area of system mnnageme.nt is logging and archiving ofevents. This is illustrated by a·case history when
the system performance during nom1aJiy slow activity time at night was poor. Further probing the system resources
indicaled that the system was busy with processes beiilg executed from outside the institution. The system bad been
"compromised;" i.e., had been broken into. The intruder could manipulate the normal system resource tools so as to
bide lhe intruder programs. The intruder was fiuaUy discovered from the archival system log.

Solutions to the total IT services are currently being offered by commercial \<endors. We wiU discuss them along
with network and system management t·ools and systems in Part ill oft he book. We will present here a high-level
view of some of the aJternale perspeclives of1 he broad aspects of net work mnnagemenL

I. U. I. Network Mllnll~eruent J'crspccti••c

Domains: ll1e ·network managemenl overvi.e w given so mr in the chapter can be perceived as managemem of a
domain. The domain can ·be any of a selected group of parameters having common attributes. Thus, a geographical
domain refers 10 1be subdivisions of a large geogrophical region. For example, In lndia 1be u:lecommunictllion
administration is div.ided into circles, and e<tch circle maintains its own telecommunication network.

Another classification of a· domain can be based on v-endo:r products. Thus. we could have different vendors'
management systems managing their respective products. A third perspective of .looking at domains can be &om
the technology perspective. For example, IP-based products, telecommunication products, broadband
communicmion products, and digital ttansport ·p roducts such as SDH cot~d each define a domain managed by a
sepamte NMS, as well as a different administrative group.

Protocols: Netwo rk management can be perceived from the protocol used 1omanage the network such as lotemel-
based SNMP and OS!-based Co mmon Mnnagemenl lnfurmation ProtocoVCommoo Management information
Service Element (CM!P/CMJSE). Tmffic nse of various protocols at each promcollayer can be monimred.

Network and Transmission Technologies: An end-to-end network system could be viewed as comprising multiple
network technologies traversing different transmission media and carrying information in different ttansmission
modes, each managed from a different network management perspective. Thus, an end-te>-end communication.
which can be represented as a logical circuit, could be made up of network elements comprising IP-based routers
and ATM-based switches. h can tmvcrse globally through coaxial cable in an access network, wireless
transmission over continents, fiber optic cable over land on a WAN, ru1d twisted copper wire at home. The
transmission mode could be digital IDM, or ATM, or a broadband access mode. An integrated NMS is used to
manage end-to-end availability of a circuit that deploys multivcndor and multitcchnology network ele.ments.

1.1 L.2. Sen •ic.e M:tnagement Pen~t>ective

The network is used to provide service to customers and consequently what needs to be managed are the· services.
The real concern of service providers is more about seme management. Providing quality of service to satisfY the
customers' needs requires network managemem. However, while network management. focuses on the physica.l
network, se.r vlce management focuses on services offured over I he network and those services meeting customer
needs and satisfaction. Various qual icy of service (QoS) parameters u.re defined and an SLA is reached between tbe
service provider and the cuslomer. There are several OSSs that provide different types of service management.

Communication services can be offered as public switched network services, lotemel. services, virtual private
network, real-time interactive audio and video services, and others too numerous to list. Computing services are
offured to clients using applications nmning on servers. These servers and applications running on them need to be
managed centrally by the service provider or enterprise that owns them. This management is also known as
enterprise management. It monitors the health of system resources, a.s well as the applications that run on them.
There are managed service offering.~ available to manage multiple enterprise networks from a common
management facility.

1. 11 .3. OSS Per.> pcctive

While the EMS. NMS, and enterprise management >')'stem are designed to manage the network and network
resources, OSSs support !he operation of network nnd service management systems. [o Section 1.9 we described
the supporting functions of networking needed to provide communication services as operations, administration,
maintenance. and provisioning (OAMP).

Provisioning System: The logical and physical network has to be provisioned to provide the. des.ired service to U1e
customer. An OSS, provisioning management system, does this function us.ing several other OSSs such as the
inventory mllJlage.ment system, the service order system, and the element and NMSs, Provisioning management
includes. circuit provisioning, service provisioning, llJld network provisioning.

Inventory Management System includes inventory of equipment and facilities. We can generalize equipment ns
active components forming nodes ofa n~-twork and facilities as passive components linking the .nodes.

Customer Relations Management (CRM) operation support system manages complaints reported by the C\L~tomers.
A proactive approach to CRM is the service provider calling the customer on detecting a service outage indicated
byNMS.

Trouble Ticket and Work Force Management manages the troubles detected by the NMS and generates work order
in the Work Force Management System. Various OSSs help with the remote testing, either on-demand or
automated, in installation and maintenance.

IP Telecommunication Application Management: The traditional analog services of voice nncl video are now
o.ffered as digital services. Such services as voice-over-IP lllld video-over-lP applications require· not only
management of data, but also coooectioo management. Sessions that are equivalent ro a circuit need to be
established and managed.

1.11.4. e-Busin~.~s Manag~men t

The <>business management and privacy requirements are associated witb EKOmmerce applications. This Includes
application management in Internet retaiJ acli vi ties, ns well as bllJlking automated ieller machines.

I.J 2. NMS P la t form

NMSs and tools are available in various platforms-hardware and operating system. Popubr higb·end systems are
housed on UNIX-based servers. Low-end NMSs run either on Windows or Linux-based platforms.

Most big)H)nd NMSs are equipped witb remote client capability and can be accessed either via Java client or Web
browser. Client platforms are e ither Windows or UNIX based.

Common troubleshooting and monitoring of network element parameters could be done by using sinnple
networking and network management tools. These are part ofTCPIIP stack. For example, nerwor.k connectivity
could be tested us ing ping and lraceroute commands in UNlX and tracert in Microsoft Windows. We will discuss
NMSs and tools in detail in Chapter 9.

1. 13. Current Status and Futu re of Network Ma nagl'mcot

Current NMSs are based on SNMP protoool. Most commercial network components have embedded SNMP
agents. Because of the universality of the IP, transport of management infOrmation for SNMP managemem, which
is TCP/IP-based, is automatically resolved. tn addition, most of the popular host-operating systems come with
TCP!fP protO<:OI suite and thus are amenable to SNMP management

Current NMSs, however, suffer &o m several limitations. One of the limitations of SNMP-based management
system is that values of man.aged objects should be defined as scalar values. OSI-based management protocol,
CMrP. is object oriented. However, it bas not been successful due to the· complexity of specifications of managed
objects and the limitation of large memoty in computer systems in the past. Another limitation of SNMP-based
management is that it is a poU-based system. ln otber words, NMS polls each agent as to its status, or fur any other
data that il needs for network management. Only a small set of transactions is initiated by a managemem agent to
an NMS as alarms. To detect a mult quickly. o r to obtain. good. statistics, more frequent polling of agents needs to
be done by the NMS, which adds to network traffic overhead. There is an alternative solution to this problem,
which is deployment of remote monitors as discussed in Chapter 8.

Some of the above constraints in SNMP-based management have been overcome by emerging advanced network
management discussed in Chapter 16. Object-oriented techno logy has reached a matured stage, and the hardware
capacily to handle object-oriented stacks Is now commercially available. Titus, object-oriented oetwork
management is being reconsidered. Tltis bas po tential application in Telecommunications Management. Net~\Wk
d.iscussed in Chapter 10. Network managemem systems are currently built with object-orierued protO<:Ois and
schema, sucb as Common Object Request Broker Architecture (CORBA) protocol and Extended Markup
Language (XML) schema.

An active network, which is tbe direction of nexi generation network, would include embedded network
management applications. Besides the advancement of research and development in network maMgemeot io
standards, protO<:Ols, methodo logy, and new technology, there is considerable a.otivity in management applications.
which form the topic of Chapter II. Of particular significance are event correlation technology in fault
management., and secured nel\vork and communication in security management.

With the proliferation of till} Internet, secured network and communicatio.n has beoome extremely impo11ant.
Existing management standards do not go far enough in this. However, security management has taken on the role
of a special topic in network management. Topics of high interest in this field are ftrcwalls that cstabHsh secure
netwo rks and cryptography that as.~ure secure oommunication.

IT itself is exploding and gives rise to new challenges for expanding the horizon of network managemen.t.
Transport of voice, video, and data is integrated in broadband multimedia services. Broadband multimedia service
is based on ATM, IP, and MPLS in a WAN and several emerging ae<:ess technologies such as HFC, Asymmetric
Digital Subscn'ber Loop (ADSL), and fixed and tnobl.le wireless. Quality of service in integrated services is
important. Managing these new service offerings forms the conte.nt ofl'art IV.

Another re-emerging technology Jbr network management is the wireless technology. T his is being widely
deployed for WAN, mobile. broadband access. and home networks. Much work on standardization of management
ofthis technology needs to be done in this area.

S umma ry
We presented in this chapter an overview of data and telecommunication netwo rk;, as well as converged networks
and how these networks are managed. Till} telephone network was shown as a model to be followed in
accomp.lishing a reliable, dependable, and quality data communication netwo(k. We elCplained the difference
between data communicat·k>n and tek:communicatk>n networks, although th.is distinction is fast disappearing.
Desktop processors and LAN technology have contributed to tbe client~rver distributed computing environment,
which has changed Lhc fltture direction of dam communication. We briefly talked aboutlhc lnternel and intranel in
t01lay's environment. Adoption of standards bas played a significant prut in the popularity ofthe Internet. OSI and
CPs play an important part in data communication today. We also treated difficulties associated with rea.J.time and
non-real-time management of different segments of broadband networks and services. We bave presented ,o;ome
practical day-to-day experiences ofnet~vork managers, including "war stories" to make us realize the importance. of
network management. We saw a bird' s-eye view of oetwor.k maoagcme01 and described how network components
and networks are managed by network mafll!gement systems. We e:.1ended the concept of network management to
managing networks and systems and all of IT services. The future direction of IT management is undergoing
changes due to advancements in software and IT. Possible futuro directions in network management technology
were addressed nt the end of the chapter.

Exercises

Note for Problems 1--4: n is important that a network administrator be familiar with both the protocols employed in
the network and the tools with wltioh its operation may be investigated. There are several tools that are
fundamental lOr administration of an lP network; the after used ones are ping, nslookup, and traceroute. These
commands sbould be available on UNIX platli>rms. You may get the syntax o f their usage by logging into a UNIX
system and accessing the 011-line manna) by invoking the command man commandna·me. Similar tools or
commands are available in Windows 95/NT machines (ping, iracert, nslookup either buill in or external software)
connected to the Internet. Problems 1- 7 are intended to familiarize you with exploring a nerwork. You should be
able to do these exercises using the commonly available networking tools and on the Inter.net using webs.ites 1iUCh
as wbois.domaintools:oom and itemic.net.

l.n d oing these exercises, if you have a problem reaching the destination host, you may use any other equivalent
destination site. lt is important fOr you to Jearn to use tools and interpret results.

1. Who Is the primary Internet service provider (ISP) In your Institution? Find another Institution served by the
same tSP by using a tr~roulll tool.

2. Educational Institutions In your state or province are networked. Discover that network by tracing the route from
your Institute or organization to other II'IStltlttes or organizations.

3. Draw the route diagram Identifying each node for the foll owing data obtain~ using a trace routing tool. Wtlat Is
the average time a packet takes to travel from noel host to netman host? noc2% traceroute
netman.cc.gatedt.edu

traceroute to netman.cc.gatech.edu ( 130.207.8.31), 30 hops max, 40 byte packets

main-rtr.gcalt.gatecb.edu ( 199.77. 147.1) 1.045 ms 1.012 ms 0.971 ms.

130.207.251.2 ( 130.207.251.2) 2.198 ms 1.404 ms 1.837 ms.

netman.cc.gatech.edu (1 30.207.8.31) 3.528 ms 1.671 ms 1.602 ms.

4. Between which two hosts on the route between your slte and www.president.l v is the largest geographic
distance probably traversed? Support your answer with evidence.

5. Ping nsl.bangi!J.net in this e~erdse. State wtlat data you gathered and how it determined your conclusion.

a Measure the percent packet loss between a host at y()ur sit.e and the machine nsl.bangla. net, and
record ihe t ime o f your measurement.
 b. Then cletennine where along the route to ns l.bangla.netlhe pac·kets are getting lost.

6. For each host on the route between your location and nsl.bangla.net (or any other foreign country), determine
the name of the administrative contacr responsible for It (use whols command from your UNIX system or from
lnternlc.net). List these names alongside the hosts. If you can' t find-an administrative contacr for some of the
hosts, then at least state what you did find.

7. You can discover the hosts In your subnetwork by usi~ the ping command with your network IP address and
host address of decimal 255. Discover all the hosts in the subnetwork that you are logged on.

8. In Problem 5, Identify the gateway from your subnetwork to others.

9. Identify the hosts In the neighboring subnetworks and draw the configuration of Interconnected subnetworks.

10. The email system Is based on dlent-server architecture. Send an email to a wrong node address (for example,
misspelling the remote node address). Explain the error message(s) that you get and the servers that you get
them from .

11. Send an email to a remote site with a wrong user id, bu t correct node address. Explain the error message(s) that
you get and the servers that you get them from. ·

1.2. Explain the decimal notation In representing the clas.ses of 1Pv4 addresses. Give an example for each class.

13. You are given a class B IP address of 145.4S.x.y for your ne~ork node. /Js a network engineer, you are· asked to
configure your network for 126 subnets. (Remember that 0 and 1 are reserved.)

a. How would you configure your address for subnets and hosts?
b. What is the maximum number ofhosls lhat eaob subnel can accommodate?

14. An IP network Is connected to a Novell IPX network via a gateway as shown. Draw the protocol layers of the
gateway in Figure 1.27.

Flgurt 1.17. Extrdsc 14

 ,_.~~ lu.~c- r---
,,.,. -
rcr
II'
1-
IPX
--
1-
t(O~ J
-
~'!J

l'lly
1-
Plw
--
I I I

15. MBI Corporation uses cc:mall, which is not Internet standard. The company also uses Novell IAN. Novell has
Internet Exchange Protocol, IPX (connectionless datagram service), as Its equivalent to Internet TCP/IP. As you
know well, most of the global email uafflc Is on the Internet with SMTP as the mall protocol. Figure 1.28 shows
the high-level configuration of the two networks connected through a gateway. Fill in the protocol layers ofthc
gateway.

Figurr 1.28. Exr•'tisr IS

SMlP c~ mn l l

- -
.
~1!.3 81)2.3
- ~tffi.'Tlltl
- ll1bcrttol

I I I I
Cabk
16. Picture a scenario where you are downloading a file from a server, located in Europe, which has an X.25 protocol
base<! on the OSI Reference Model. liS physical medium Interface Is X.21. Your client machine lsronnected to the
Internet with Ethernet as the physical medium.

a. Draw the delllils of the CXlmmuniClltions oel\vork in Figure 1.29(a) using bridges, routers, and a
gateway between lhe server and the client.

Jllgurr 1.29. E•r,..isr 16

 0_ -----~§I~N
t:tJ iW,\N~-----~
Notwork ~
Cl ..-nl S<n<:r
(a) L·oumuUltCdlh.m NUt,\•cuk. Uetwccnt.ll\111 ttnd .S..·I\-cr

ICal~>n U)'<:~ i\pp:1c:mon 1-ll)'l'f<

>po<l Luy•l\l
-
lnt o1nl.!c~n t <11111!\\ll)'

C~II\'Ct)ii.:!
i-
'Ll:•y~n
1'11lMpon U)<t<

I I'A>~•col Medium
I I l'llyS~a~l M<<hum
I
(b) C:~n1munlcmlon lki\\~CII fnd S)'I'ICI"-' via .111 l•tenncdlnJ.: S)"'<nl

b. Complete the protocol architecture in Figure I.29(b) for the intermediate gateway system.

17. In Case History 2 described In Sertlon 1.7.2, the delay In alarm indication In INMS was attributed to several
possible causes. Give an example for each of these causes.

18. ~a network engineer in a Network Operations Center, you are followl"'! up on two troubl e tid<ets. You do not
have a network management system and you have to use the basic network tools to validate the problem before
you can resolve them. Please explain what tools you would use In each case and how i t woul d validate the
customer compl aint

a. Trouble Ticket 100: C ustomer says tbat when be receives messages, the message is periodicaU y
missing soroe cbarncters.
b. Trouble Ticket 10 I: Customer in Atlanta complains that when she t·ries to log into the system
server.beadquarters.com in New York, she gets disconnected with a lime--out. However, her
c<>lleague in ber New York office "''POrtS that he is able 10 access the system.