Professional Documents
Culture Documents
Chapter 2 - Local Area Network Architectures
Chapter 2 - Local Area Network Architectures
,""*Sb.
wffi
&
rl .,.r,:rr,r
'r
Concepts Reínforcecl
OSI Model Top-Down Model
Hardware / Soffware Compatibility Protocols and Standards
Local Area Networks
Concepts Introdt¿ced
Access Methodologies Network Architectures
Physical Topologies High-Speed Network
IEEE 802 Standards Architectures
Logical Topologies
.0BJBCTIY.SS
36
The OSI Model Rer-isited J/
ii$ INTRODUCTION
Chapterlansweredseveralkeyquestions'atleastonanintroductorylevel:
o What is a local area network?
o How is a local area network implemented?
. Why are local area networks implemented?
what, how, and why of local area
Armed with a basic understanding of the
networks, this cnapter ñü, iridepth the underlying local area network
technologies,to transpar-
ur"t-rlt".t.rr"s that allow ñardware and software
able to communicate and share infor-
entlv interact. Clients and servers are that
;:"ffi;.;;;r;,h" combined components of the tocat area network
are able to offer trans-
links them. These local area netwoik components
to ciients and services due to their
parent network transmission services
adherence to standará-r-una protocols. One of
the key distinguishing char-
acteristicsofaparticularlocalaleanetworkisthenetworkarchitecture
the components of a net-
adhered to by a particular LAN' In this chapter'
work architecture *iil;; first explored, folüwed by comparative evalua-
tionsofthenumerousnetworkarchitectureseithercurrentlyavailableor
emerging into the networking marketplace'
first intr;duced in Chapter 1' will be explained in
more
The OSI model,
detailasameansbywhichtoorganizecomparativeinformationregarding
communications'
altemative network architectures aid comput"?-to-.or1}p.tter
Physical Iayer
The physical layeq, also known as layer 1, is responsible for the establish-
ment, maintenance, and termination of physical connections between com-
municating devices. These connections are sometimes referred to as
point-to-point data links. The physical layer transmits and receives a stream
of bits. There is no data recognition at the physical layer.
Specifically, the physical layer operation is controlled by protocols that
define the electrical, mechanical, and procedural specifications for data
transmission. The RS232-C specification for serial transmission is an exam-
ple of a physical layer protocol. Strictly speaking, the physical layer does not
define the specifications for connectors and cables, which are sometimes
referred to as belonging to layer 0.
Data-Link Layer
The data-link layer is responsible for providing protocols that deliver reli-
ability to upper layers for the point-to-point connections established by the
physical layer protocols. The data-link layer is of particular interest to the
study of local area networks because this is the layer in which network
architecture standards are defined. These standards are debated and estab-
lished by the IEEE (Institute of Electrical and Electronic Engineers) 802
committee and will be introduced and explained later in this chapter. The
number 802 is derived from the date of the committee's formation in 1980
(80) in the month of February (2).
The data-link layer provides the required reliability to the physical layer
transmission by organizing the bit stream into structured frames, which add
addressing and error-checking information. Additional information added
to the front of data is called a header, whereas information added to the back
of data is called a trailer. Data{ink layer protocols provide error detection,
notification, and recovery.
The data-link layer frames are built within the network interface card
installed in a computer according to the predetermined frame layout par-
ticular to the network architecture of the installed network interface card.
Network interface cards are given a unique address in a format determined
by their network architecture. These addresses are usually assigned and
preprogrammed by the NIC manufacturer. The network interface card pro-
vides the connection to the LAN, transferring any data frames that are
addressed to it from the connected network media to the computer's mem-
ory for processing.
The first two layers of the OSI model, physical and data-link, are mani-
fested as hardware (media and NICs, respectively), whereas the remaining
layers of the OSI model are all installed as software protocols.
The OSI Model Rer isitec 39
Sub-Layers
Inordertoallowtheoslmodeltomorecloselvadheretotheprotocolstruc-
tné mgE 802 committee split the
ture and operation or u rolur area netwofk,
data-link láyer into two sub-layers'
NlediaAceessControlThemediaaccesscontrolorMACsub-layerinterfaces
by protocols fhat define how the
with the physical hy;;;; t; represented
It shared local area network mediá
is to be u..érr"d by the many connected
'.4
computers. As will
fully later is this chapter' Token Ring
';"ütiflrnore use different media access
1- (IEEE 802.5) ar,d Etit"Á"i GEEE 802'3) networks num-
Jt methodologies and ;il";;"re assigned different IEEE 802 protocol
are com-
ES
bers. Uniqu" uda,",'"' assigned to
NlCs at the time of manufacture
*r"fy t"iltred to as MAC Jdd'"tt"t or MAC layer addresses'
of the data-link layer that inter-
Logical Link Control The upper sub-lay-er
á ft""*'l u''th" logical tilk-:"-T1t^t^or LLC
sub-
faces to the network Ñ"' (IEEE-8.2'2)' The LLC
layer and ir r"pr"r"''t;áü rgng 802
,'ttgi" irotocot
sub-layer also inter"f"ac"" i'utJpu'enily to the MAC sublayer protocol
Ji- to rpfiitir,g tt* dutulitrt layer into
two sub-layers
beneath it. The to
the "a"""ág"
to having u ,ir,gi", .lo**o' LlCprotocol
is that it offers transparency
and
the the upper layers and above) while allowing the MAC sub-layer
i""tlott
ork protocoltovaryindependently.Intermsoftechnotogy,thesplittingofthe
allows a givén network operating
rab- subJayers and the siirgle LI-C protocol network architectures
802 ,yt,""{ ," t"" "q""il;:"Jl *"u var;ftY of differént
fhe as embodied in network intertace
cards'
980
Transport Layer
Just as the data-link layer was responsible for providing reliability for the
physical layer, the transport layer protocols are responsible for providing
reliability for the end-to-end network layer connections. Transport layer pro-
tocols provide end-to-end error recovery and flow control. Transport layer
protocols also provide mechanisms for sequentialTy organizing multiple net-
work layer packets into a coherent message.
Tiansport layer protocols are also supplied by a given network operat-
ing system and are most often closely linked with a particular network layer
protocol. For example, NetWare uses IPX/SPX in which IPX (Internet Packet
Exchange) is the network layer protocol and SPX (Sequenced Packet
Exchange) is the transport layer protocol. Another popular transport,/net-
work protocol duo is TCP/IP in which TCP (Tiansmission Control Protocol),
the transport layer protocol, provides reliability services for IP (Internet Pro-
tocol), the network layer protocol.
Session Laver
Presentation Layer
Application Layer
The application layer, Iayer 7 of the osl model is also open to misinterpreta-
he
tion. Application-layer protocols do not include end-user application pro-
ño
grams. i{ather, they include utilities and network-based services that
ro-
íupport end-user application programs. Some people include network oper-
ati'ni systems in this categoty. Stti.tty speaking, the best examp^les of appli-
et- X.400 and X.500. X.400 is an open
catiJn-iayer protocols ute th"OSI protocols
systems proio.ot that offers inteioperability between different e-mail pro-
aL-
gru*, utt¿ X.SOO offers e-mail direótory synchronization among different e-
!er írail systems. DNS, Domain Name Service, which is an Internet protocol that
ket or domain name to a specific IP address, is
resolrrls a computer's common
ket protocol.
also considered an application-1ayer
ret-
Figure 2-1 offeÁ a conceptual view of the osl model and summarizes
01),
)ro- many of the previous comments.
The previous discussion highlighted the roles of the various osl model layer
protlcois in a communication session between two networked computers'
rño
!Lót i{ow the various protocol layers actually interact with each other to enable
are an end-to-end communication session is highlighted in Figure 2-2.
r.ilar As illustrated in Figure 2-2, a dafa message emelges from a client front
nr-i- end program and procéeds down the protocol stack of the network operat-
rate, ing systám installed in the client PC in a process known as encapsulation.
rion, Ea"ch'successive layer of the OSI model adds a header according to the syn-
tax of the protocol which occupies that layer. In the case .of.the data-link
rent, layer, both a header and trailer are added. The bit stream is finally passed
aláng the shared media that connects the two computing devices. This is
ram- an
impJrtant point. Although the OSI model may seem that given lay-
fo-1mply
in u prótocol stack ta'ik direcily to each other on different computers, the
"r,
fact is tñat the computers ut" ot-tÍy physically connected by the media, and
that is the only layer which talks directly between computers'
When thé fuil bit stream arrives at the destination server, the reverse
process of encapsulation, de-encapsulation, takes place. In this manner,
:.ica- each
:lica- !.r.c"rsirr" layÉr of the OSI moáel removes headers and/or trailers and
iered processes the data that were passed to it_from the corresponding layer
proir-
- --l iol o. the source client. Once the server has processed the client's reques: :::
j
i üat data in the server back-end engine application, the whole process is rer-er==
)r are and the requested data will bé encapsulated by the server's protocoL .:::'
-'
:¡cols transmitted over the communications media, and de-encapsula:e; f ü'e
42 Chapter TWo Local Area Network Architectures
DATA ENABLING
LAYER USER AFPLICATION
'.'. FORN,lAT TECHNOLOGY
--
7 APPLICATION Provides common seruices to user
applications.
+ X.400 E-N¡AlL interoperabjlity o
specification =
o
.> X.500 E-l\ilAlL directory c
synchronization spec¡f ication c
o
+Strictly speaking, does nof
include user applications ,9
C
l
E
6 PRESENTATION Provides presentation seruices for E
network communicat¡ons. o
o
'+ Encryption
C
'=
.>Code translation (ASCll io !
É
EBCDTC) ! .9
C
r>Text compression l
/Vofto be contused with o
r> Graphical User lnterfaces(GUls) o
!
E o u
c C
E
5 SESSION Establishes, mainla¡ns, terminates o
o o sessrons Distr¡buted
! o applications,
! F
.E
zo lnteractive, real- middleware, or ro
=
.9. time dialogue network ¡t
o operating
between 2 user
o systems.
o nodes
o
4 TRANSPORT Assures reliability of end-to-end
-g c messages Network
netwofk connections- o Operating
c o Systems
.9 c Asembles
I c
o packets inio
messages.
o
:
C
3 NETWORK Establishes, maintains, and
o
packets Network
terminates end-to'end network 5 Operating
connections. o o Systems.
Embedded within
=
o o rames.
f
z
E
c
U
NIC DRIVERS
APPLICATION
' ''iviñ -
7 APPLICATION
trtrtr r
LAYEF
f^--¡s¡
fiL¿atal
.. r dára
L - r-D¡t¡
DATA
PRESENTATION
6
LAYER
PRESENTATION
illilllii LAYER
il
i1
ll
IJ
iol
F
1<l
J
l5
UJ
IL
l<t
tol
\z
l.'l
t¡
lt
n
|.r
i'l
LAYER
il
I Str""t
Bit
V
to the client
client PC's protocol stack before being ultimately delivered
first place'
front-end application which requested the data in the
. access methodology
. logical topology
. physical topology
Access Methodology
Realizing that more than one user is likely to be sending requests onto the
shared local area network media at any one time, the need for some way to
contfol which users get to put their messages onto the network and when
should be obvious. If the media is to be shared by numerous PC users, there
must be some way to control access by multiple users to that media. These
media-sharing methods are properly known as access methodologies. Shar-
ing the mediá is an important concept to Local Area Networks, which are
sometimes referred to as Media Sharing LANs.
Logically speaking, there are really only two philosophies for controlling
access io u Ánur"d media. An analogy of access to a crowded freeway can
provide a vivid illustration of access methodology choices.
csil,tvcD one philosophy says, "Let's just let everyone onto the media
whenever they want ut-t.l if t*o users access the media at the exact same split
second, we'll work it out somehow." Or ... in the analogy"'" Who needs
stoplightsl If we have a few collisions, we'Il work it out later!"
Tñe access methodology based on this model is known as Carrier Sense
Multiple Access with Collision Detection, or CSMA/CD for short. A clearer
understanding of how this access methodology works can be achieved if the
name of this access methodology is examined one phrase at a time'
Carrier Sense: the PC wishing to put data onto the shared media listens
to the nettr.ork to see if any other users are "on the line" by trying to sense a
The Local Area Network Architectwe \foc¿r {5
transmission is sensed' then
neutral electrical signal known as a carrier' If no
;ü"ii+f" Access,, illor* anyone onto.the media without further permission
a free line and access the
,"q"lr"'J. rinally, if two user'pcs should both sense,,Collision Detection,, lets
media at the same instant, a collision occurs and
theuserPCsknowthattheirdatawerenotdeliveredandcontrolsretrans-
mission in such *";';i; avoid further data collisions' Another possible
"
factorleadingtodata,collisionsisthepropagationdetay,whichisthetimeii
takes a signal from a source PC to réach
atestination PC. Because of this
for.a workstation to sense that there is no
frofugu,ion delay, it is possible
sisnal on the shared when in fact another distant workstation has
^éaiu, PC.
;ililil ,1g""i,n"t has not yet reached the carrier sensing
"
Intheeventofacollision,thestationthatfirsidetectsacollisionsends
Each workstation,
out a special lamming signat to all attached workstations. is preset to
card in the workstation'
;;;.;" p.e.irely the"niwork interface thus reducing the like-
wait a random amount of time before retransmitting,
continue to occut the
lihood of reoccurring collisions. If successive collisións
random time-out interval is doubled'
CSMA/CDlsou,,io.,stymostefficientwithrelativelylittlecontention
PCs to access the network
for network ,"ro.rr.ár. ine abiiity to allow user
requesting and granting reduces over-
easily without excessive permission
neiwork usage rates' As usage
head and increases p"'fá'*utte at lower
of data collisions and retransmis-
increases, however, ih" ir-r.reus"d number
,lor-t, .ur', rt"gatively effect overall network performance'
the media *l
Token Passing The second philosophy of T:th"q?l:gl: T::iT::
access
:ffi"iffi:1 Ji;;;ó;;;;i; á"'¿ifuá'"" l';: I"::i:::,:?:
I give vou the m:sic tk"ifl1lI"^"-l:'
Lhe
-to il;;;#;?": pJ'ii'"á", Jnd onlv rnJ analogv would,be the con-
'ren ;; il; ;;,á tn" ,nur"i mediai' nigirwav
áti"et must wait at a stoplight and
,o*rr**uyr in which
trolled access ramps
somehow immediately get to á0 mph in order to -:lq:,**tth" Tlf]i;-
^'^-
.ese
ilffi;;;ile ejcn pc user has 10Ó% "f ll:"1Y::i*H
xaf-
;;;;rffi ;i;' "'i',1'"",r-'^t
;;iu- r"q.,"rt, and transf ers by. insisrtig,,1^,r,l"'^,u^c.
ñJ p ás,e,sing. u,p".if i. u:k"
are ""1
;::":JJJfi:'#iiir *i,n"; I :,91 l*:), :t
ffi:'ñff;;;;'k;". rhe token is generatéd in^the firsi place bv, t1a desig-
Df-
hg nated pC known ur,f* Ji,r" monito"r and passed among PCs until one
^,^ ^ PC
aan
would like to access the network'
changes the token sta-
At that point, the requesting PC seizes the token'
onto the network' and doesn't
tus from free to b"d;;; its"data frame
successfullr-'
edia its data were delivered
release the token untí ii is assured that rvork-
split confirmed by the destination
Successful delivery of the data frame is the fra¡:.'.
¿eds successful receipt of
station setting frame status flags to indicate the
around the ring to sen'.'ll]-,:
and continuir-rg to tor*urd the Jriginal frame
flags set to " i..:--
ense
lC. Uf"" ,"ceipt of ,n" "rigi""l fiame with frame status the sending FC :e='=
nation ad.dress ,".og',i,"d, í'u^" copied successfully"'
¡ the ii. After the se:-:'-. FC
the token status f.r;];t to free and releases
next pC, r,r'hich n-ra1.' =r:: ir ir-:.f
releases the token, it ;; ;;;á"Jalong to the
riens
the free token or Pass it along'
nie a
46 Chapter TWo Local Area Network Architectures
+ Station D:
Station A:
+ Grabs 24 bit token and sets T- + Receives frame and identifies itself as
bit from 1 to 0 (tree to busy) the destination address
+ Puts data frame on network. + Moves data from f rame into memory
+ Sets destination:Station D + Changes frame status flags to confirm
.> is in transmit mode. delivery
Station B:
+ Receives frame and reads tts + is powered off so network
destination address traffic flows through
+ Sees destination is Station D unprocessed.
and copies the received frame + is in bypass mode.
to the output link
+ '.s in l¡sten mode. ABBREVIATED FRAME LEGEND
T busy status
A address recognized
C copied successfully
DATA transmitted data
I ao source address
DA destination address
rt
Logical ToPoloE
l Once a data message is onto a shared
LAN media which is connected to
rl to be determined is how that mes-
numetous workstations, the next thing
IS
sagewillbepassedfromworkstationtoworkstationuntilitultimately
:I
r)
,"Í.n", its inünded destination workstation. The particular.message-pass-
known as a networkarchitec-
ing methodology employed is more properly logical topologies has to
¿"r.tiue
ture,s logicat toporog:;"i;;ü;:"á'to
S-
in a PC user's wastebasket'
do with the best *uy'io put out a"iire
rg
method of delivering data is known
Scquential The first logical topology or
?r,
e11
as sequential. in a ,-qÍ-"ii"Liogiár
topology, also known as a ring logical
rk, PC (oí'node) to another' Each node
topology, clata are passed f'onione
nt daia packet to determine if this par-
examines the destinatlon uddr"r, of the
ticularpacketismeantforit'Ifthedatarverenotmeanttobedeliveredat
thisnode,thedata"'"p"*"aalongtothenextnodeinthelogicalring.
This is th" b.r.k"t ü;tg"d" logiial topoiogy
*:11"9:f puttlng out a fire
is filled by o_1e pC user and
in a pc user,s wastebasklt. A bricket ot*uüi
rlassedtotheneighboringPCuser.Thatuserdeterminesifhis/herwaste-
do""' the flames with the bucket of water'
i"ri"itt ""'irt"l?llü1n? t""' U"tftet along to the next user in the logical
Otherwise, tn" .,r"'iá*"tif-t"
ring.
Ph,vsical ToPolog.v
be ohvsicallv con¡ected tt' r'-':l' -:=r
Finally, the clients and servers must
d" Í.**¿'bv the shar'; 1-':r .'i
u..oráir-tg to some configuration and
48 Chapter Trvo Local Area Network Architectures
Ring The ring topology suffers from a similar Achilles'heel. Each PC con-
,-recied via a ring topológy is actually an active part of the ring, passing
data
packets in a seqlential pattern around the ring. If one of the PCs dies, or a
network adapter card Áauunctions, the "sequence" is broken, the token is
lost, and the network is down. In addition, any cable breaks bring down the
entire network.
star The star physical topology avoids these two aforementioned potential
pitfalls Uy e-pioyittg some type of central management device. Depending
tn the t.,"'t*o.t aichitecture and sophistication of the device, it may be called
a hub, a wiring center, a concentrator, a MAU (Multiple Access unit), a
r."p"ui"r, o. u ,ñit.hing hub. All of these devices will be studied later in the
text. ny isolating each ÉC or node on its own leg or segment of the network,
ur-ry .rádu o, .ub"l" failure only affects that leg, whereas the remainder
of the
network continues to function normally'
Because all network data in a star topology ale going through this one
central location, it makes a marvelous spot to add system monitoring, secu-
rity or management capabilities. Conversely, since all network data are
going througñ tnt one central location, it makes a marvelous networking
no-no known as a single point of failure. The good news is: any node can be
lost and the network"wiil be fine. The bad news is: lose the hub and the
whole network goes down.
As we will iee shortly in the study of hubs, vendors have risen to the
occasi.on, offering such reÍiability extras as redundant power supplies,
dual
buses, and "hot swappable" intérface cards. Figure 2-4 highlights the differ-
ences between these physical topologies'
il$ NET\I'ORKARCHITECTURES
Ethernet
ither
!'ers.
Bus topology
Network
,ffi
ru#d
logy,
Lls to
r¡'ork.
Star toPologY
r"*
- con-
Sdala
i/ora
ken is
rn the
¡tential
rrding
: called
Init), a
r in the
rtlvork,
r of the
ihis one
€/ secu-
lata are
Figure 2-4 LAN Physical Topology Choices
n-orking
b canbe
the
and the 802'3 are conflicting standards'
strictly speaking, Ethernet and IEEE net-
term Ethern¿t ir .om"tonly used
to refer to any IEEE s02'3 compliant
rr to the
work. Differences b";;; th" *t standards
*itt U" outlined shortly'
lies, dual
he differ-
Functionalif
. Access MethodologY: CSMA/CD
. LogicalToPologY:Broadcast
. Physical Topology: Traditional$-Bus;
Currently-Most often Star
StandardsThefirstEthernetstandard'wasdevelopedbyDigital,Intel'an.] to
t \{etcalfe i'¡g8l, and was known as DIX i'0' sometimes referre'l
Xerox Corporation
Although s
50 Chapter Tn'o Local Area \etu'ork Architectures
as Ethemet I. This standard was superseded in 1982by DIX 2.0, the current
Ethernet standard, also known as Ethernet II. The frame layouts for Ethernet
II and IEEE 802.3 are illustrated in Figure 2-5'
As illustrated in Figure 2-5, both Ethernet II and IEEE 802.3 frames can
vary from 64 to 151,8 octets in length.
The Ethernet II frame starts with a preamble of eight octets. The pur-
pose of the preamble is to alert and synchronize the Ethernet net-
work interface card to the incoming data.
The destination and source addresses are each six octets long and are
also known as MAC layer addresses. These addresses are Perma-
nently burned into the ROM (Read Only Memory) of the Ethernet II
netwbrk interface card at the time of manufacture. The first three
octets of the address identify the manufacturer of the network inter-
face card and are assigned by the IEEE. The last three octets ale
assigned by the manufacturer producing unique MAC layer
addresses for all Ethernet network interface cards.
The type field identifies which network protocols are embedded
within the data field. For example, if the data field contained Net-
work IPX/SPX protocols, the type field would have a value of 8137
(hexadecimal) and if the data field contained TCP/IP protocols, the
type field would contain a value of 0800 (hexadecimal). These type
,rul.rer are assigned by the IEEE. The type field is important in order
to enable multiple protocols to be handled by a single network inter-
face card which enables multiple protocol stacks to be loaded in a
Network Architech:res 51
E01.3 frame, it must refer to the header of the IEEE 802.2Logtcal Link Con-
trol (LLC) data unit. Figure 2-6 illustrates the fields contained in the IEEE
802.2 data unit.
6 Octets 6 Octets
1 Octet i 1 Octet I
4 ,,Base,,referstobasebandtransmission,meaningJhattheentirebandlt-i'lth
of the media is ¿"rrotáálo one data channel.
T!^etT" stands for Tu Lste*1Pak'
themediaofchoice.Anotherimportantdistinctionofl0BaseTLsthalitsrec-
54 Chapter Trr'o Local Area Network Architectures
ifies the use of a star topology with all Ethernet LAN segments connected to
a centralized wiring hub. Other Ethernet standards and associated media are
listed in Figure 2-7.
Maximum
Standard Popular Name Speed Media Segment Length
10Base5 Frozen yellow i0 Mbps Thick coaxial 500 meters
garden hose cable (RG-11)
(.405 in. diameter)
10Base2 ThinNet, 10 Mbps Thin coaxial 185 meters
CheaperNet cable (RG-58)
l0BaseT 10BaseT, 10 Mbps Unshielded 100 meters
twisted pair twisted pair
Ethernet
l0BaseF Fiber Ethernet 10 Mbps Multimode Described by IEEE
FOIRL (Fiber- fiber optic 802.1j-1993
Optic Inter cable standard
Repeater Link) (1000 meters)
Token Ring
Origins The credit for the first Token Ring network architecture has been
attributed to Olaf Soderblum, who proposed such a network in 1969. IB\t
:t has been the driving force behind the standardization and adoption of Token
l- Ring with a prototype in IBM's lab in Zuriclr., Switzerland, serving as a
5 model for the eventual IEEE 802.5 standard.
te
)- Funetionalit"v
a
5S
Access Methodology: Token Passing
It- Logical Topology: Sequential
d
Physical Topology: Traditionally-Ring, Currently-Most often Star
or
)n Standards Unlike IEEE 802.3 Ethernet networks that have a speed of 10
Lta Mbps specified as part of the iEEE standard, the IEEE 802.5 Token Ring stan-
to dard does not include a speed specification. IBM, the leading advocate of the
ne Token Ring network architecture, has specified Token Ring network archi-
in tectures that operate at 4 Mbps and 16 Mbps.
As mentioned earlier in the discussion of the token passing access
methodology, the token is actually a24-bit formatted data packet and is illus-
trated in Figure 2-8 along with the IEEE 802.5 Token Ring MAC sub-layer
frame layout.
The IEEE 802.5 token frame layout consists of the following fields:
. The starting delimiter fieid alerts the Token Ring network interface
card installed in a workstation that a Token Ring frame is approach-
ing. Notice that both the token frame and the MAC sub-layer frame
both start with the starting delimiter.
o Once the Access Control field is received, the workstation can distin-
guish between tokens and MAC sub-layer frames. If the token bit
within the Access Control field (see Figure 2-8) is set to 0, then the
received frame represents a free token, in which case the access con-
EE trol field would be immediately followed by an ending delimiter.
The workstation is welcome to receive the full token frame and
change the token bit from 0 to 1 to indicate a busy token. The
received starting delimiter field plus the access control field with the
T (Token) bit now set to 1 form the first two fields of an IEEE 802.r
MAC sub-layer data frame, allowing the sending workstation to iust
append address information, data, and the remaining fields in the
data frame layout and transmit this frame onto the ring.
¡ If the token bit on the received frame was set to 1, then the ne-.r -;-:
is the frame control field which indicates whether this l¡ame . --:::- :-'
data or is a special network management frame.
56 Chapter TWo Local Area Network Architectures
<+8bits=+
3 bits i '1
bit I 1 bit 3 bits
1 brt 1 bi1
Fígn'e 2-8 IEEE 802.5 Token Ring and MAC Sub-layer Frame Layout
. Following the frame control field are the destination and source
address fields. The receiving network interface card would read the
destination address to determine if it was the frame's intended recip-
ient. If it was, then the workstation will read the rest of the frame into
memory. If not, then the NIC will simply pass the rest of the bits of
Network Architectures f/
the frame along the ring without transferring them to the worksta-
tion's memory. No snooPing allowed.
The routing information field is used with devices known as soufce
routing bráges which are able to link together multiple Token Ring
LANsISourie route bridging and other LAN-Io-LAN connectivity
options will be discussed in Chapter 14'
The IEEE 802.2 header fields are used in an identical manner as
they
are used with IEEE 802.3 Ethernet MAC sub-layer frames. similarly,
IEEE302.2SNAPisalsosupportedwithintheIEEE302.5MACsub.
layer frame.
The data field contains data in the form of embedded upper-level
protocols if this is a d.ata ftame, and network management informa-
tion if this is a network management frame as indicated by the frame
control field. The daia field dóes nothave a fixed maximum length
as
the tasks
all other workstations are designated. standby monitors' Among
thatcanbeperformedbytheactivemonitorarethefollowing:
by
. Removes frames from the ring which have not been removed
their sending workstations.
. Regenerates lost or damaged tokens
that it
. Provides a special 2l-bilbuffet if the physical ring t:.t?.:-u11
token' For
does not have enough delay or latenry to hold the 24-bit
;In Sharper Focus" entitled "Token Ring
more information, sée the
and Timing."
r Controls the master clock
on this
. Makes sure that there is only one designated active monitor
ring.
Network Architectures 59
it
)l
rg
Note that in the IEEE 802.3 transmission the least significant bit (BlT 0) ls transmitted last
l]s
IEEE 802.5 Transmission
EE
ff,
5S-
€t- Note that in the IEEE 802.5 transmission the least significant bii (BlT 0) is fansmitted first.
ter a
TOKEN RING AND TINTING
net-
rhile In order for the Token Ring network architecture to operate correctly, t}']te 24-,
ture. h Sharper bit token must circulate colntinuously even if no workstations are in need of
and Flrcus transmitting data. Therefore, the entire Token Ring network must possess
oken enough delay or latency to hold the entire 24-bit token. This latency or
ñ-een requiied deláy can be computed by dividing the tength of the token (24 bits)
ds. by the ring's iransmissiorr speed (a Mbps) yielding a required latencv of 6
Ái.rot".oids. The next question is how far can an electrical signal travel in 6
microseconds? The ur-ts*"t to that question will depend on the media
through which the signal is traveling, with different media possessing diJter-
; net- ent pópagation velocities. For example, unshielded twisted pair has a plt-:-
e 2-9, agaiion rre"locity of .59 times the speed of light, denoted as "c." The sPee'j u'i
Tu'o Local Area Netrvork Architectures
: ilo%?,",:"ff"*.,
whelmingmarketshareofEthernet.Performanceisnotsignificantlydiffgr-
is possible,
ur,¿ iiteroperability between the two network architectures
"r.,t Ethernet/Token Ring bridges provide trans-
although challenges dá exist.
will be
p"r""i?,.top"rubility between the two network architectures and
detailed in ChaPter L3.
ARCNBT
3t- a popular local area
ra tit'la:l:uf ¡l:tlrr,,''- ARCNet (Attached Resources Computer Network) was
Inc' It
he Il,,Shfiper network architecture that was originally developed by Datapoint'
offered 2.5 Mbps transmission speed and used a token-passing
Focus access
rat topology over
u üroud"ust logical topology, and a star physical
tu- -"inoaoftgy, to connect IBM3270
RG-62 coaxial cable. Since nÓ-OZ is the sáme cable used
terminals to cluster controllers, ARCNet was often installed
in downsized
could be reused. ARCNet was never stan-
IBM installations where the cable
has been largely replaced by Ethernet and Token
d.ardízedby the IEEE and
Ring network architectures.
ork
{-ar.
met FDDI
Eth- network
sof Origins Fiber Distributed Data Interface (FDDI) is a 100 Mbps
and architecturethatwasfirstspecifiedinlgS4bytheANSI(AmericanNational
rom Standardslnstitute)subcommitteeentitledX3Tg.5.Itisimportanttonote
that FDDI is not an IEEE standard. However, FDDI does
support IEEE 802'2
rt 10
i"gi."f Link Control protocols offering it transparent interoperability to
:ken
IEñE compliant upper-layer protocols (layers 3-7)'
most
;and Iunclionalin
5 afe
CCESS
. Access Methodology: Modified Token Passing
ue to LogicalToPologY:sequential
[orver
'
r-hich
. Physical Topology: Dual Counter-Rotating Rings
sults.
Data Inter-
reater Built-in Reliabilif and Longer Distanee FDDI (Fiber Distributed
high
If one f"*¡ t"ppfi", ,tot only a gieat deal (100 Mbps) of bandwidth'-but also aproto-
st dif- J"j."" fi reliability and"security while adhering to-standards-based
cois not associated with or promoted by any particulut l:"d91 . .
ls urunune
r inte- FDDI's reliability comes not only from the fiber itself, whrch
(Radio Frequencl lnter-
longer to both EMI (Electro Magnetic Interfbrence) and RFI
through
i"rurr."¡, brri u. additional degree of reliability is achieved
th-e
nbrace
-
design of the physical topology of FDDI'. . l
hernet foDt', p'nysi.ut toptlogt is comprised of not one' but t\\-o' s€p;re:É
e ovel- rirrg, arorrttá inl.n ¿uiu *áí" simulfaneously in opposite
direcd¡:-' Ü:¿
62 Chapter Two Local A¡ea \etl ork Architectures
.i.g is the primary data ring, whereas the other is a secondary or backup
data ring to be used only in the case of the failure of the primary nng or an
attached workstation. Although both rings are attached to a single hub or
concentrator, a single point of failure remains in the hub while achieving
redundancy in the network media. Figure 2-10 illustrates some of the key
features of the pppl network architecture and technology, while Figure 2-11
more specifically illustrates the self-healing capabilities of the dual counter-
rotating rings network architecture of FDDI.
m ádditiot-r to speed and reliability, distance is another key feature of an
FDDI LAN. Up to 5Ó0 nodes at 2 km apart canbe linked to an FDDI network.
The total media can stretch for a total circumference of up to 200 km (125
miles) if repeaters are used at least every 2 km. This increased distance capa-
bility makés FDDI an excellent choice as a high-speed backbone network for
camPus environments.
Another positive attribute of FDDI, illustrated in Figure 2-10, ts its abil-
ity to interopérate easily with IEEE 802.3 10 Mbps Ethernet networks. In this
outer loop
transmission
direction
SAS controller
O...''-.-
../..,;::'
FDDI br¡dge
(accesses single Ü
loop onlV)
l\,4icrocomputer
Minicomputer
l\¡icrocomPuter
l\¡inicomputer
\::trl*r3"'-\
adapters
n
uier
\
in order to
!vay, a business d.oes not have to scraP its entire existingnetwork
bridge is the
a piece of it to 100 Mbps fópt. An FDDI to Ethernet
"pir"a"
rp""ifi. technology employed in such a setup'
' Th" technolo"gy involíed with FDDi nelwork architectures is
similar in
and is illustrated in Figure,2-
function to that oIbtn", network architectures
l0.PCs,workstations,minicomputefs,ormainframesthatwishtoaccessthe
FDDI LAN must be equippediith either internal FDDI network
adapter
cards or external FDDI controllers'
One way that some network managers cut down on
FDDI's cost while
is to connect only !o_o1e of
still benefiting from the 100 Mbps banáwidth
is sometimes called SAS, or
FDDI,s two fiúer rings. This type of connection
to DAS, or Dual Attachment Sta-
Single Attachment Stations, ui oppos"d-are
tion"s, in which both FDDI ringi accessed. Obviously, if a device is
by
attached to only one FDDI rini it forgoes the reliability afforded the
¡First,becauseofthegreatpotentialdistancesonanFDDILAN'it
let a
was impractical to tuln "fÁ'e" tokens into "busy" tokens and
single siation monopolize that token until it had received
confirma-
Token Ring,
tion of successful delive'y of its data. Instead, unlike
byte from 0 to 1 and
which just flipped the T bít in the access control
the token from the
appenáed u áituframe, FDDI physically removes
,i.g ur-ra transmits a full data fráme' Upon
of transmis-
T-p1"t1o"
sioi, it immediately releases a new token. Recall that Token Ring
waited until the transmitted frame returned before releasing the
token,Collisionsarestillavoided,sinceonlyonestationcanhavethe
the
free token at a time, and stations cannot put data messages onto
network without a token'
Network Architectu¡e> 65
Backbone
DI
rd
ia
üy
rle
ta-
ro-
lth
rut-
rLd
Workgroup
rire
ous
AN
;ail-
ion.
AfIS-
mto
oles,
¡'ork
toa
S,
rgh-
|ta1
; are
rte-
le
Figure 2-13 Alternative Applications of the FDDI Network Architecture
Ittr
re via
to ing a few (fewer than 20) PCs or Workstations that require high
ces
¡an¿wi¿ttr communication with each other. Multimedia worksta-
I ring
tions, engineering workstations, or cAD/CAM workstations are
0re
good exa'mples of higt-r bandwidth workstations. As "Power users"
s and
iurn increaéingly toward high-bandwidth Graphical User Interfaces
ner is
(GUI), this consiituency's bándwidth requirements will rise as $-e11"
3. High Bandwidth sub-workGroup connections-In some cases.
¡ory is ottiy t*o or three devices, perhaps servers, need high bandr'':;d':
¡nnect-
68 Chapter Two Local Area Network Archltectures
AfterallofthesepositivethingsaresaidaboutFDDI,surelytheremust
architecture' Chief among
¡" ,o*"tr,ir-rg negative about thisiAN network remains to
th" p?.", although how long price wiil be a negative
";;;;;i, architecture, in order {or a
be seen. As with any other ,f,ur"d media"network network
pc to access an FDDI laÑ, lt must be equipped with an FDDI
with the "lower" priced
uaupt", .urd. These .urd, ,ur-,g" ttoÁ $SOOtá $ibOO, of the two
to and use only one
FDDI network adapter .u.dr"ubl" to attach at
the average Ethernetcard $25
FDDI data rings. Compu." tr"r"t" ftices with t\1t'ryce'
twice
tt $200;;Jth'" ut"'ug" Token r..tJng card at approximately
increases in the FDDI tech-
A; ñ;i;;s in"popularity aid competiiion
that they
-",k""t, p,i.",'*ill unááubted!' f;11, although it is doubtful a nega-
""r"gy itself is seen as
will ever reach Ethernet price levels. The fiber media canbe packed
ti"" f;;;;;;;" u, *áfr. ef,trough fiber is lightweight and
of glass and can break. Also, con-
more densely than copper wire, it ii made
necting, terminating, una tfil.-it-tg iiber-optic
iuut"t require special tools and
traininq'Theseobstatf"'á"-fCo'""o''"and'atleástinsomecases'the
of the unknown.
may be nothing more than the fear
"f"J;?;rb;;
FDDI'S FUTURE
Occasionallyinthefieldofdatacommunicationsandnetworking,onetech-
Managerial that might be cheaper'
nology is eclipsed or replaced-by a newerlechnology
Perspecth e be the fate of fOpt' At one time' FDDI
,.:¡,r*:t¡r¡¡¡r.::1i¡;
easier to work with, o. totft' fnit may
to turn to when Ethernet and
was the only network architecture aíternative
.'iait::r:11*:trl::11::r:.
100BaseT
offering 100 Mbps
Ethernet standards
lOOBaseT rePresents a familY of Fast The
methodology'
performance and adhering to the CSMA/CD
access
High-SpeedNetworkArchitec¡res 69
DII\IETER
TIMING TSSUAS AND 100BaseT NET\\'ORK
CSMA/CD-based network architecture'
ech- Collisions are a fact of life with any
lPer/ Ir Thetimerequiredforagivenworkstationtodetectacollisionisknownasslot
occut the transmitting station
DDI For:us time and is measured ;i uitr. \{hen collisions
the affected frame canbe retransmitted'
:and mustbe notified,r tn"."riirirn so that
; will However,thiscollision"otifitationandretransmissionmustoccurbeforethe
100BaseT is 512 bits or
peed slot time has expired. il" ,tot time for both loBaseT and
¡r{or- unof;:'ro"ed as l.BaseT.In order
ill be of 100BaseT is obviously 10 times as fast
tobecertainthatcollisionnotificationsarereceivedbyl00BaseTnetwork-
attached"-workstationsbeforetheirconstantslottimeexpires,themaximum
to the increase in net-
network diameter must be reduced proportionately
network diameter shrinks from 2'500
work speed. As a result, the maximum
meters to 210 meters'
TeehnoloSMostofthel00BaseTNICsarecalledl.0/100NICs,u'hichmea:rs
or 100Base! but not simult¿r-L-
that they are abte ," ,";;;;;i;ther l0BaseT
than quality loBaseT orLlr- cards a--it-;"--
ously. These .ur¿, .o,iit'igntty *o'"
)MbPs
¡-. The
IT
100VG-AnvLN
IEEE 802.3 and IEEE i]02.5 Support 1OOvG-AnyLAN',s abiliiy to support IEEE
802.3 and IEEE 802.5 frame types and networks more specifically means:
1 0BaseT
shared media
1 OOVG
(token ring)
1 00VG Token
Ring Nlos
Network A¡chite'-:';-
Eigure 2-15 IEEE 802.3, IEEE 802'5, and 100VG-AnyLAN
72 Chapter Trvo Local Area Network Architectures
Demand Priorif Aceess Perhaps the most unique aspect of the 100vc-Aty-
LAN network architecture is ihe Demand Priority Access access methodol-
ogy,alsoknownasDPMA(DemandPriorityMediaAccess).Thisunique
access methodology eliminated the collisions and retransmissions
that are
characteristic of Ethernet and the token rotation delays of Token
Ring' Key
points of this access methodology are as follows:
As illustrated in Figure 2-1.6, tkte central hub, also known as the controlling
',,$$iátper hub or the root hub, controls the access to the network, whereas all lower-
Focus level hubs maintain communication with attached workstations and main-
tain address tables identifying which workstations are attached to which
hub to
ports. when a workstation requests permission of its locally attached
ioad a message onto the netuiork, tñat request for permissron is passed up
until it reachés the designated controlling hub'
through the híerarchy of hubs
The pérmission granted in response to this request totransmit.onto the
net-
to the initially requesting
-ork i, passed áown through the hub hierarchy
workstaiion. Remember, a liaximum of only four hubs or repeaters can lie
between any two end nodes or workstations'
High-Speed Network A¡chitectu¡e: 73
ói ooNTROLLI.NG.'húb' tt
ú'
5, + Bequesi
:l---'---{ á
-:r
r!
Permi$ion
to Transmit
STEP #3. UPÉérlayer
hub passes..réquegt to
lsree *a. ooáró¡inf f' STFP'#5.'j:rcrintiolli¡gi i
lub. the¡':óof li.HUts #?
.o.
< - ->
controlling hub. o
tt,
Instructions
..
iñ.,Roúrid.Robiñ {aahi¿n to Seruice
(J Requested
I -. r.. r,: HUB.#2:r, Transmission
Sree +, , t-o"át,nuU
passes request to
I
;
i/\
I
-.riÚi:r*s I
e
€
r/ ,r=,
" ilIl"ff--l
of
requests Permission
transmit
I
local hub to
l
| ¿a
--;-rl
¡
rn
@@@E
|--:i--,r-Él
退= -
ts-
ll' Figure 2-16 Cascading Hubs and the Round Robin Poiling Scheme
rty
ol Technolog,y The implementation of a 10OvG-AnyLAN requires compliant
ta- NICs, driver software, and hubs. 10OvG-AnyLAN NICs cost between $260
Iri- and $315, whereas hubs cost about $230 per port. For ease of migration most
10/100 Ethernet 1O0vG-AnyLAN NICs are available for less than $300. It is
rfo- important to determine whether this 1O0VG-AnyLAN t9:T9!8y w1ll !e
transporting Ethernet frames or Token Ring frames. Specific NICs and hubs
must'be puichased for each of the two transported frame tyqgt, with Ether-
ret- net 1O0Vt-AnyLAN technology being more readily available than Token
lnS
Ring 1O0VG-AíyLAN technology. the lndlsputable fact is that regardless of
coriparative tec'hnical merits, 1óbBaset has won the Fast Ethernet network
architecture battle and investment in 1OOVG-AnyLAN technology should be
carefully scrutinized.
[ing
tl'eI- Isochronous Ethernet
rah-
hich Although the ability to transpoft time-sensitive traffic such as voice, video,
rb to or mult]media is one of the ádvantages of 1gSVG-AnyLAN, other network
dup architectures such as Isochronous Ethernet, also known as Iso-Ethernet, can
hub. also effectively transport such traffic, although not at 100 Mbps perfor-
:net- mance. Details of the iso-Ethernet network architecture are contained in ihe
sting IEEE g02.9a standard, which is officially known as Isochronous Etherrtei
m lie Integrated Services. The term isochronous refers to any signaling sr-stem rr-
whiáh a11 connections or circuits are synchronized using a single cotr1-''rl--i:
74 Chapter Trvo Local Area \etl-ork Architectu¡es
mechanism allows ti:l^Y,t::l:
clocking reference. This common clocking to streamrng or
,;if*?""tanteed delivery times, *Ttl are very important
time-señsitive traffic such as voice and video'
.Al0MbpslsDNPchannelisreservedforEthernettrafficandis
channel
In fact' this P
completely coÁpatible wl!!-f0e1eT Ethernet'
can be T NICs' allowing network managers to selec-
"t"iüí08;*
tively or grui,'uffy migrate to Iso-Etheinet offering
the multimedia
require it'
.upuUititié, to only thoie workstations that
,A6.l.44MbpslsDNCchannelisreservedforstreamingtime-sensi-
tive traffic such as multimedia applications'
.9664KbpslsDNBchannels,whichcarrytheactualmultimediatraf-
B channels as needed up
fic. Applications are able to aggregate th-ese
to the 6-1'44MbPs limit'
.l64KbpslsDNDchannelwhichisusedformanagementtaskssuch
as call control and signaling'
Trre6.l44MbpsCchannelwhichcarriesthemultimediatrafficusesthe
same8KHzclockingsignalasthecommerciallsDNWANservicesoffered
interface between the Iso-
by long_distance cañieñ, enabling a transparent
clocking signal" is the
Ethernet f-ex u',íWÁÑ ,"g-áts. Thii "same
derivationofthetermlso(sameas)chronous(timing).
Networkdiameterofanlso-Ethernetnetworkislimitedtol00metels
fromthemostdistantLANworkstationtotheWANinterface,Iso.Ethernet
rwisted Pair, allowing it to
runs over t*o p* ol cutegory 3 or 5 unshielded
operate orr"t network wiring in many cases'
Iso-Ethernet ,-t"Euo'tt' operate ii tht"" different
"iistit-tg service modes:
for Ethernet
L. l0BaseT Mode-Uses only the 10 Mbps P channel
traffic.
P chalnel for Ethernet
2. Multi-service Mode-Uses both the 10 Mbps
and the 6.144 Mbps C channel for video/multimedia'
(248 x 64 Kbps chan-
3. All Isochronous Mode-Uses all 16'1'44Mbps
óf isochronous band-
nels) for t*^*i"g protocols' This amount
width will suppoitieal-time video or voice distribution'
High-speedNetworkArchitecf-ir<: :5
n
Y
rS
et lsochronous Ethernet
lsochronous Ethernet
Access Unit (AU) with
to Access Unit (AU) with
integrated ISDN-WAN
integrated ISDN-WAN ISDN
ISDN interface
interface WAN link
WAN link
10 N,4bps
1 0 N/lbps
Ethernet traffic
Ethernet traff¡c
et
MicrocomPuters and Server Microcomputers and Server
equiPPed with 10Base-T equipped with 10Base-T
network interface cards. network interf ace cards.
1 0NIbps Ethernet
'10BaseT
the L00 Mbps standard is that it supports a standard for virtual LANs
(Chap-
ter 3) known as IEEE 802.1q that will allow Ethernet flames to be encapsu-
lated within Token Ring irames, an important characteristic in those
organizations that must support both network architectures. Although
100
Mtps Token Ring NICs, ,*it.h"t, and router modules may be available'
.emember that cJmpatible driver software must be available to allow any
network operating system to be able to communicate with these devices' To
date, high spe"a iotótr ring has generated little interest in comparison
to 100
Mbps Ethernet alternatives.
to-serv.er cffilt"tritJ.rpr",.
instarled in the computer is both
sending
NIc
áperating svstem and
and receiving data ,Jii;;'";ü;";.;;i"hreaded technol-
to take full advantage of this
network operating 'y;i;;;'" required systems and network oper-
ogy. Exampl", of *ultithreáded operating
"'ct-' g'tt and 4'1' as well as most
afing system, ur" witdo;Ñi
oSlz'ÑetWaré
?- varietiesofUNIX'FullduplexEthernethasgatheredSufl;ie*llterestfrom
to warrant the for-
and.user"communities
u-
the networking technology vendor standards t"tt duplex
to propose I:l or avail-
SC
mation of the IEEE 802'á; committee
JO
Ethernet. Fullduplex;;;;i";y
i"t'o "itnJr under development
Ring (32
(200 Mbps)'|"il Duplex Token
Ie,
able in Full Duplex F;EÁ;tfi"t
ny ñ;;, ruf puplex FDDI (200 Mbps) varieties'
and
To
00
Gigabit Ethernet
also known as 1000Base-X' is an
Nehlork Arehitecture Gigabit Ethernet' by
gtn"rn*i?nut was sta¡dardized as the IEEE 802.32 standard
uosrade to Fast following configurations:
the ff"tff;; ;ü;;;1d;;' rh" stándard defined the
hes
fiber-
-d L- ¡ 1000Base-sX: uses short wavelength (850 lTgl:t"tt)-laser
building cabling on a
nce optic media, p"r;;rú-used for ñorizontal
Lere given floor
.od- (1300 nanometers) laser fiber-
for . 1000Base-LX: Uses long wavelength appli-
hi"gh-sPeed campus backbone
'een optic media, p;;;'iit ;'"d fo'
¡.ted cations
cable and transceivers for dis-
¡lta-
' 1000Base-CX: Uses copper twinaxial a data
rnet tances of only 25 meters' used
primarily to link servers within
iitnm a wiring closet
ru11 center or high-speed network devices
ruch
1000Base-TX: Expected to be
rat'fied tr 1999' this standard would
han ' four pair of Categorr 5
rorts atlow gigaUii tt't-t""-t*t to travel over meters
of 100
ould unshielded twisted pair at a distance
:ated
SpecificsofthegigabitEthernetstandardarelistedinFigure2-79,IlshouLc
recommended distance o\/er FDDI->:-'
re
'and-
be pointed out that the maximum
.rplex *.riti*o¿" fiber-optic cable is only 220 meters'
r. for
78 Chapter Trvo Local ,\rea \eirr rrr\ -\:chiie¡illres
Fib er Fiber
Standard TyP" Diameter Bandwidth Range
Fibrc Channel
While gigabit Ethernet runs at 1000 Mbps (or 1 G-bps), an alternative net-
*ork ui.Éitecture known as fibre channel (ANSI standard X3T9.3) has been
High-SpeedNetworkArchitelturs 79
LAN-Based ATM
that
is a switched network technology
ATM (Asynchronous Transfer Mode) per sec-
r'o'.'-t 2-5 wp-t to.several gigabits
has been defined ,t ,P;;;;;"gi'-tg and
available for both workstations
ond. Network interiace cardi aie with
ATM-based computers-to communicate
servers. However, i"';;;";;;
known as LAN emulation must
be
non-ATM-bur*a to*"pJü"' u ptot*ss
implemented.LAN"*t'lutio"ándotherAlM-basednetworkarchitectures
and will be described in more
fall into the realm t;';;;tptte networking
order for aJpiications to take full
advantage
detail in Chapter 3. M;;;"'"r, in )afrnn awate"'Although desktop
of ATM's ,p""a u''aTJu*"'"t't"y gustfe
as animation and stock trad-
AIM has been implem"r1t"¿1r", slch industries are
high-spe"d rt:t*:ll,arcllitectures
S
ing, many of tf.,e prwüusiy rn"ttioned how-
t-
more tikely,o o* ,*püLJ'.;;;
;;;ilal,#*u networks. With that said,
hrgh-speed backbone networks'
s ever, ATM is a fairly popular choice for I
5
l-
NETWORK ARCHITECTURES
IN A NUTSHEII
rS
stlnt{Rr
In order to properlv anaTyze and design local interest is the data link layer that serves as the
area netrt-orks, it is absolutely essential to home of the IEEE LAN standards and is sub-
have a thorough understanding of the OSI divided into the MAC and LLC sub-layers for
model and its ionstituent lavers. Of particular that purpose. The processes of encapsulation
Ke.v Terms t1
RErlElV QUBSTI0NS
1. \\l-rat is the importance of the OSI model to Define the relationship between the two data-
local area network analysis and design? link layer sub-layers.
2. \\tLat is a protocol? 8. What does the introduction of data-línk layer
3. \\hat is the relationship between protocols sub-layers offer in terms of increased interop-
and the OSI model? erability options?
4. !\hat is the or-erall purpose of the physical 9. In general, what are the purposes of the
lal'er? header and trailer added to datalink layer
5. \\Trat are the major differences between a frames?
point-to-point 1i¡k and an end-to-end link? 10. Where are data-link layer frames built and
6. \A4ry is the data lirk laver of particular inter- why is this an appropriate place?
est to LAN netrvork architectures?
-\ct1\-1¡'Es $3
to Toker'
the network 31. FIow are timing issues significant
11. What is the relationship between Ring networks?
and Token
laYer?
layer and the data-link 32. bi-ff?rentiate between Ethernetat various trai-
between-the trans-
12. tínuitl af't" relationship ñirlg1; t"t-t of performance
laYer?
t.riitu.t and the network more interest to
fic levels'
I 1 ít;;iJ ih" session tayer of \Arhat advantages does
FDDI offer over Ether-
/ 33.
clieít / server informa tion systems
as to the
net and Token Ring?
14. Name at least two misconcáptions
34. ;;;i; ine s"lln"áti"s Powers of FDDI'
ity'
i.,i*to,"tutio" of layer f unctional
encapsuta-
?6 ii,#:;ffii''*"tti'i*í'v negative attributes?
primárv ules of the FDDI
15. Briefiy explain the purpose of JO. wh;;;;; ir,.
tion / de-encaPsula tion' network architecture?
that make up any and disadvantages
16. Wná, ur" theihree elements 37. wh;i;;" the advantages
network architecture? to FDDI?
and token of CDDI as oPPosed
17. éo*pu" and contrast CSMA/CD 38. Wttu. i, the a'diantage of dual lo*]ngt, -^
Passing as access
methodologies'
of col- c;;o;* the advantáges and disadvantages
18. ítnli,l'" different poten"tial causes
*" ^é; of 10'0BaseT and FDDI'
defined for
lislons in Ethernet networks? 40. l.r"rlU" the three standards
release mechanism
79. What does the early token 100BaseT'
of buying 10/100
accomPlish? 41. Wn"i it the advantage
20. What ictuallY is a token? NICs?
a logical and a protocol'
,"t'. iry-rt"i it trt" dirr"t""t" between 42. il"t.tiU" demand priority
PhYsical toPologv? 43,. Ho* dot' demaná priority ptoloto]¡ljil"
broadcast and does not get Perma-
22. 5iír","ntiuit ¡"iíeen the af'tu, fo* priority traffic
sequential logical topologies' r -' t^J"t out of network access?
bus' star' and rlng
23. nif'fer"'-'tiut"ietween the ¿;;ó";" the advantages and disadvantages
44. """,iY
Phvsical toPologies' .iioósuttrand looVG-AnY LAN'
Ethernet II and IEFE uat'u'"tiug"s aní disadvantages
of
24. b'1filt';i" u"ñ"tt" 45. Compare tt""
Ethernet' ánd 1O0VG-AnyLAN'
802.3 2 Isochronous Ethernet
25. Wi"r^, i' the relationship between IEEE 802 46. What are some unique
attributes of Isochro-
and iEEE 802'3? *h;ti compared to other high-
nous Ethernet
IEEE 802'2 and Ether-
26. óiiie'"ntiat" between ,o""d n"t*otk architectures?
net SNAP' 47. Ñhut do"t isochronous mean?
between Isochro-
27. What is a protocol discriminator various media-spe-
;Á:,. p"r.ttu" the relationship
,.t,. Oiff"t"t-'tiát" between the of Ethernet'
nous Ethernet and ISDN'
cific alternative configurations 49. d;;" the advantages and disadvantages
t"tllt''e ihara^cteristics of a Token
29. What are tt"'e of gigabit Ethernet?
disadvantages
Ring network architecture? in a 50. Wf;;? "t" the advantages and
30. üf-t?iit role of the active monitor of high-sPeed Token Ring?
'f-tt network?
Token Ring
Iata-
. i,:'., .1
ACTMTITS I:. ,
fLrP-
From the previous activity'
determj¡e u'hic't'
2.
or bulletin board con- do not conform rr el! :'-
1, ' -r- a' presentation
Preoare
' laYer model' As
:ru#:í;;p*ái"r'
sisting ot an emPty \)slT of the oSI model'
u-#o."rrt iáYer
operating sys-
local area networKs or network
protocois in J. bitin" an alternative network comm'!r:-:r-
--;: :
l:d tems are encountered' place the
il#;;;,"*imodel to the oSI mol¿.
the ProPer laYers of the OSt mooei'
84 Chapter Two Local Area Network Architectures
why the new model is more effective than the reasons for their choice and add your own
OSI model. analysis of the results.
4. Choose a particular protocol stack and outline 10 Survey schools or businesses that employ
the frame layouts for each protocol in each FDDI network architectures. Gather informa-
layer of the OSI model. Be sure to indicate tion regarding motivation, instaiiation date,
reiationships betr,veen protocols as to which satisfaition, problems, and the outcome of a
protocols are encapsulated by which other similar decision on network architecture
protocols. made today.
{ Investigate the IEEE 802'4 Token Bus stan- 11. Investigate the availability and cost of tech-
dard. Report on its history, implementation, nology for the three standards defined for
available technology, current status, and an 100BaseT. Analyze Your results.
explanation of this current status. t2. Investigate the availability and cost of tech-
6. Survey the local area network implementa-
nology for 100VG-AnyLAN varieties' Ana-
tions in your school or business. Report on lyze your results.
the physical topologies found. Explain your 13. Compare the availability and cost of 100BaseT
results. technology and 10OVG-AnyLAN technology'
7. Investigate the daisy chain physical topology'
Analyze your results.
Is it truly a unique physical topology or a 14. Investigate the cost and availability of
variation o{ one of the three primary physical Isochronous Ethernet technology. Analyze
topologies? your results.
8. Conduct a sLrrvey of Ethernet networks in 15. Investigate the gigabit Ethernet and high-
your school or business. What is the media of speed Token Ring markets. Which market has
choice in each installation? Why was each more technology to offer? What are the statis-
media chosen in each situation? tics on market size for each technology?
9. Survey schools or companies that have
Analyze and present Your results.
installed Token Ring networks' Report on the
Crsu Sn'nt
Goodhve Token Ring, H_gllg Plh_._tngl
......-...........:....
\takine the move from token The company hadbeen con- has cost the comPany more
ring to Ethemei? lÉ so, go as far sidering moving awaY from its than $l million.
as vou ca¡r as fast as You can. 800-user token-ring network Given that the network can
That's the advice from for several years. The time now speed traffic along five
Equitable Liie Assurance seemed right when the com- to 50 times faster, Equitable
Socieh-'s IT ,ctaff, which pany decided to uPgrade its figures the project will PaY off
recentlY iinished moving the desktop and selver Platforms in increased ProductivitY and
comPanv's sen-ice organiza- and consolidate its service the ability to suPPort
tion from a sha¡ed 1/16]ll{' organization at a new site here. expanded workflow and
bit/sec token-ring en\-iron- The migration project/ imaging applications.
ment to a su'itched 10//100M which involved grouPing Because it was moving to a
bitlsec Ethemet network. three service centers into one, new site, Equitable ditched
Quesüon: S5
Business Case StudY
be cost
3. How could the network migration
Activities justified?
. Comolete a top-down model for this case by
' ;;tdt;lacts'from the case and placing tl:- i" Application
on the new
i;;;;il;ter or the toP-down model' After r.^ Ho* many desktoPs are supPorted
n m odel' anaty e
g'.o*pleted the top-dow
7 network?
nurrln
requirements
ofe u.t¿ ¿ltuif those instances where Data
from upper layers to both to
*1." .f"-fy passed down 1. What level of bandwidth is available
solutions
can io*"t iuy"á of the model and where the desktoP and on the backbone?
five io those requitements were passed "P Í?*
nbie lower layers to upper layers of
*" t:O"ll^^. Network
Equitable
What network architectures did
the
l. Detail any questions about the case that may 1.
¡ off answers are not clearly move from and to?
;*";;í"; for which
and
stated ln the article' 2. Wflui *"t" the relative merits of the oid a¡"1
port new network architectures?
and Business 3. How was the migration Planned?
u'orkstatici'-' ¡: -:e
4:,. er" tl't"." any Token Ring
;to a this Particuiar time? new network?
migration?
ched l. tf'hr'at was the cost of the network
86 Chapter TWo Local Area Nehvork Architectures
5. Is Token Ring to Ethemet trame conversion 2. What types of functionality does the chosen
required? switch offer?
6. How is the frame conversion handled? 3. What different t)?es of computing platforms
7. Differentiate between the functionality offered and operating systems must be supported?
by the different classes of switches and routers.
Technology
1. What type of switch is used in the network
architecture?