`CHIJPTEH 5 How a Router Works
`
`Routing algorithms use a metric to determine the suitability of one
`path over another. The metric can be several different things such as
`the path length, the actual cost of sending the packets over a certain
`route, or the reliability of a particular route between the sending and
`receiving computers.
`
`For example, RIP, a distance Vector routing protocol, uses /yap count
`as its metric. A hop is the movement of the packets from one router
`to another router. If two paths are available to get the packets from
`one location to another, RIP will choose the most desirable path
`based on the smallest number of hop counts. Figure 5.5 shows an
`internetwork where two paths are possible for the routing of packets
`between the sending and receiving computers. Because Route A
`requires only one hop, it is considered the optimum route for the
`packets.
`
`Metrle : Hop Count
`
`Route A :1 HOP
`
`Packets take
`F10uis A
`
`$00194
`
`
`
`_....._...___.....__.=e.»:r;:2rr:-r~c:m=;—:::;wwrr;-wag:-,...‘.“,‘;‘rx...,(...,.::T:.w:v~v._mm-zz-zrx~_-2:-;—v::,m+::.mm.‘.~_,.....
`
`;~_ updatéioutifigitables
`
`Route B = 2 Hops
`
`98
`
`
`
`
`
`-~4.SrF.7i3.®§~.'.z7~Jrq.h>£-r:r.':‘L\‘..'vv:::a1m:a:vn:«
`
`PA RT l
`
`Types of Routing Protocols W€lli!ll’TEli 5
`
`The problem with routing protocols that use only one metric (such
`as hop count) is that they become very single minded in their pursuit
`of the best route for a particular set of packets. RIB for example,
`doesn’t take the speed or reliability of the lines into account when it
`chooses the best path, just the number of hops. So, as shown in
`Figure 5.5, even though Route A is the best path according to the
`number of hops (and RIP), you are forced to route your packets over
`a slower line (the 56-lcilobit leased line). This line is not only slow, it
`also costs you money. Route B is actually over wire that the company
`owns (part of the network infrastructure) and is actually a faster
`medium (fast Ethernet at 10OMbps). However, when you use a
`routing protocol that uses hop count as the metric it will choose
`Route A.
`
`To overcome the lack of flexibility provided by hop count as a met—
`ric, several other routing protocols that use more sophisticated met-
`rics are available. For example, the Interior Gateway Routing
`Protocol (IGRP) is another distanee—vector routing protocol that can
`actually use 1 to 25S metrics depending on the number set by the
`network administrator. These metrics can include bandwidth (the
`capacity of the lines involved), load (the amount of traffic already
`being handled by :1 particular router in the path), and communica-
`tion cost (packets are sent along the least expensive route). VVhen
`several routing metrics are used together to choose the path for
`packets, a much more sophisticated determination is made. For
`example, in the case of Figure 5.5 a routing protocol that uses met-
`rics other than hop counts (such as communication cost) would
`choose the route with more hops but less cost to move the packets to
`their destination.
`
`lypes st éfieutlteg Pteteeels
`Real—world internetworks (particularly those for an entire enterprise)
`will consist of several routers that provide the mechanism for moving
`packets between the various subnets found on the network. To move
`packets efficiently it’s not uncommon to divide several connected
`routers into subsets of the intcrnetwork. A subset containing several
`member routers is referred to as an rzrezz. When several areas are
`grouped into a higher—level subset, this organizational level is called a
`romiing domain.
`
`99
`
`
`
`PART! Networking Overview =
`CHAPTER 5 How a Router Works
`
`Figure 5.6 shows an internetwork divided into areas. Each area is
`terminated by a high—end router called a border router (or core
`router as mentioned in the sidebar). The two border routers are con-
`nected to each other, which, in effect, connects the two routing
`domains (or autonomous sys :e111s on an IP internetwork).
`
`VS-"‘L’é
`5
`Area Border Router
`
`A
`
`i
`~. 9,314,‘ :
`Area Border
`Rouler
`
`The fact that internetworks can be divided into logical groupings
`such as routing domains (or autonomous systems) gives rise to two
`different kinds of routing protocols: routing protocolsthat provide
`the routing of packets between routers in a routing domain and rout-
`ing protocols that provide the routing of packets between routing
`domains.
`
`netted by a highenénd
`'o'uter‘calIed .'9'border_ '_~
`routéfnr mré ramer.
`V
`
`I71.te1‘ior Gateway Protorolr (IGPS) provide the routing of packets within
`the routing domain. IGPS such as RIP or IGRP would be configured
`on each of the routers in the router domain.
`
`100
`
`
`
`PART!
`
`Types of Routing Protocols CHAPTER 5
`
`Protocols that move data between the routing domains are called
`Exterior Gateway Protocol: (EGPS). Examples of 4 GPs are Border
`Gateway Protocol (BGP) and Exterior Gateway Protocol (EGP).
`
`interior Gateway Protocols
`
`The Interior Gateway Protocols consist of distance-vector and link-
`state routing protocols. Several different IGPs are available and vary
`on the number of metrics used to determine optimum routing paths.
`The oldest IGP is the Routing Information Protocol and is discussed
`in the following section, along with some of the other commonly
`used IGPS.
`
`Routing Information Protocol
`
`Routing Injirrm/ttion Protocol (RIP) is a dista.nce—vector, IP—routing pro-
`tocol that uses hop count as its metric. And although it is the oldest
`IGP, RIP is still in use.
`
`RIP sends out a routing update message every 30 seconds (by Cisco
`default), which consists of the router’s entire routing table. RIP uses
`the User Datagram Protocol——-UDI’—(part: of the TCP/IP stack) as
`the encapsulation method for the sending of routing advertisements.
`
`RIP is limited, however, in that the maximum number of hops that it
`will allow for the routing of specific packets is 15 . This means that
`RIP is fine for smaller, homogenous internetworks, but doesn’t pro-
`vide the metric flexibility needed on larger networks.
`SEE ALSO
`
`" For infarvmztion on co/zjigtmfig RIP on /1 Cisco router; see page 202.
`
`Interior Gateway Routing Protocol
`
`T/ye Interior G/zteway Routing Protocol (IGRP) was developed by Cisco
`in the 19803. IGRP is a distance—vector routing protocol.
`
`IGRP uses a composite metric that takes into account several vari-
`ables; it also overcomes certain limitations of RIP, such as the hop
`count metric and the inability of RIP to route packets on networks
`that require more than 15 hops.
`
`.
`4.
`. ,
`;. suiinetmasksvvnl
`’ cussed in Chapter 1 1‘;
`'
`‘A
`I
`'
`
`:
`
`’
`
`I
`
`101
`
`
`
`PART I Networking Overview
`
`CHAPTER 5 How a Router Works
`
`IGRP (when compared to RIP) also employs a longer time period
`between routing updates and uses a more efficient format for the
`update packets that are passed between routers. IGRP also supports
`the use of autmzomom symzvm‘ (similar to the areas discussed earlier in
`the chapter), so routers running IGRP can be sequestered into
`domains where the router Lraffic in a particular domain remains
`local. This cuts down on the amount of router broadcast communi-
`
`cations using up valuable bandwidth throughout the entire internet—
`work.
`
`IGRPE metric consists of a composite that takes into consideration
`bandwidth, delay, load, and reliability when determining the best
`route for data moving from a sending node to a particular destina-
`tion node. The following list describes how each of these network
`parameters is viewed by IGRP when the routing algorithm is used to
`build or update a router’s routing table:
`
`a Bzmzlwidt/1 is the capacity of a particular interface in kilobits. A
`serial interface may have a bandwidth of 100,000 ldlobits (this
`would be a serial interface Connected to an ATM switch, which
`
`typically supplies this amount of bandwidth). Unfortunately, the
`bandwidth of a particular interface isn’t measured dynamically
`(measuring the actual bandwidth available at a particular time)
`but set statically by the network administrator using the band-
`width command. Nlore about setting serial interfaces will be dis-
`cussed in Chapter 15, “Configuring WAN Protocols.”
`
`Delay is Lhe amount of time it takes to move a packet fiom the
`interface to the intended destination. Delay is measured in
`microseconds and is a static figure set by the network adminis-
`trator using the delay command. Several delays have been com-
`puted for common interfaces such as Fast Ethernet and IBM
`Token Ring. For example, the delay for a Fast Ethernet interface
`is 100 microseconds.
`
`Reliability is the ratio of expected—to—received lceepalives on a
`particular router interface. (Keeprzlives are messages sent by net-
`work devices to tell other network devices, such as routers, that
`the link between them still exists.) Reliability is measured
`
`dynamically and is shown as a fraction when the show interface
`command is used on the router. For example, the fraction
`2 55/2 5 5 represents a 100 % reliable link.
`
`102
`
`
`
`PART I
`
`Types of Routing Protocois CHAPTER 5
`
`a Land is the current amount of data traffie on a particular inter-
`face. Load is measured dynamically and is represented as a frac-
`tion of 255 . For example, 1/25 5 would be an interface with a
`minimal amount of traffic, whereas 250/255 would be a fairly
`congested interface. Load can be viewed on the router using the
`show interface Command.
`
`As you can see, IGRP takes a lot of information into consideration
`when it uses its algorithm to update a r0uter’s routing table. It is
`often implemented in larger internetworks where RIP would be inef~
`fectual.
`
`SEE ALSO
`
`> Fo7‘i1zfm/mtian an to/zfig-ming IGRP on /I Cisco rontez; rec page 204.
`
`Open Shortest Path First Routing Protocol
`Open S/7o1'tcrtPmt}.1 First (OSPF) is a link—state protocol developed by
`the Internet Engineering Task Force (IETF) as a replacement for
`RIP. Basically, OSPF uses a shortest path first algorithm that enables
`it to compute the shortest path from source to destination when it
`determines the route for a specific group of packets.
`
`OSPF employs the Hello Protocol as the mechanism by which
`routers identify their neighbors. Hello packet intervals can be config-
`ured for each interface on the router that is using OSPF (the default
`is every 10 seconds). The command for adjusting the Hello Interval
`is ip ospf hell-3—interval.
`
`OSPF routing networks can also take full advantage of the ‘
`autonomous systems feature on large IP networks (also discussed in
`this chapter as areas and domains), which keeps link—state advertise-
`ment of member routers local to a particular autonomous system.
`Area border routers are used to connect the Various autonomous sys-
`tem areas into one internetwork.
`'
`
`Exterior Gateway Protocols
`
`As mentioned earlier, Exterior Gateway Protocols (EGPs) are used
`to route traffic between autonomous systems (routing domains).
`Border Gateway Protocol (BGP) is a commonly used routing protocol
`for inter—domain routing. It is the standard EGP for the Internet.
`
`103
`
`
`
`PART I Networking Overview
`
`CHAPTER 5 How a Router Works
`
`BGP handles the routing between two or more routers that serve as
`the border routers for particular autonomous systems. These border
`routers are also referred to as core routers. Basically, these core routers
`serve as neighbors and share routing table information with each
`other. This enables the Core routers to build a list of all the paths to
`a particular network.
`
`BGP uses a single metric to determine the best route to a particular
`network. Each link is assigned an arbitrary number that specifies the
`degree of preference for that link. The preference degree number for
`a particular link is assigned by the network administrator.
`
`104
`
`
`
`105
`
`
`
`106
`
`
`
`107
`
`
`
`PAHT II
`
`Router Interfaces CHAPTER 6
`
`Cisco routers such as those in the 2500 Series family basically are
`off—the—shelf routers that come with a predetermined number of
`LAN, WAN, and serial ports. Higher—end routers like tl1e Ciseo
`4500 are modular and actually contain empty slots that can be filled
`with several different interface cards.
`
`Not only are different interface cards available (such as LAN versus
`WAN), but the number of ports on the card can also be selected. For
`example, one of the three empty slots on the 4500 router can be
`filled with an Ethernet card that contains six Ethernet ports. Figure
`6.2 shows the Cisco ConfigMal{er hardware configuration screen for
`the Cisco 4500 router (you will work with ConfigMaker in Chapter
`16). Three slots are available (shown on the right of the screen) and
`can be filled with several different cards (listed on the left of the
`screen).
`
`‘
`
`2 Said, l8A:)|nc/S3
`4lSD_H Dmfl-l.S/ll
`3150}: with sm‘
`1 T1/ISDN
`
`Modular routers (like the 4500) designate their ports by connection
`type, followed by slot number, followed by port number. For exam-
`ple, the first Ethernet port on an Ethernet card placed in the router’s
`first slot would be designated as Ethernet 1/0 (the slot is designated
`first, followed by the port number).
`
`Viewing the interfaces (and their status) on a particular router is han-
`dled by the show interfaces command. Figure 6.3 shows the results
`of the show interfaces command on a 2505 router that has one
`Ethernet port (E0) and two serial ports (S0 and S1). The status of the
`various ports is related to whether the ports have been connected
`(physically to the internetworlr) and whether they have been config-
`ured.
`
`108
`
`
`
`PART II Router Design and Basic Configuration
`
`CHAPTER 6 Understanding Router Interfaces
`
`1-uuLu1-2>1xlu1\l intel-Iar:s:s
`Etlmrnutfi is u A
`lint:
`rutacnl in up . using hub H
`re, a dvest is HH1ll.7b3a.5Ilb3 (biz flfl1fl.7h3a.5Flh3)
`llarduarn is [an
`Internet address: is 1u.qe.1.D z§S.z<m.u.-'3
`NIH XSKGB h tau. Bl} ififlfifl Khit. DLY UESBB ueuc, 1-sly 255/255.
`luail 1/255
`arzk nut sat. kucpnlive set <18 mm)
`Encapsulac on mum,
`loop!)
`IIRP typu: HRPII. Mil’ Iinauul: -i:flD:Bfl
`Last
`inmlt mwer, output fl:flfl:BB, nut1z\|1: hung nmmr
`Laat
`I:1u¢1~1nu uf "shun interlm:-:" counter: never
`Output queue fl/40,
`I3 den: =;
`input queue 8/75, 0 drops
`5 ninutn input: 1-ate B hi I:/sun, H packets:/sac
`5 ninuta uuepue 1-eta a bit:/sac. u packet:/am:
`Ia paeluats input, II hycas_ I! no bunni-
`Received fl
`|1I'naflcant:I4 V! 1~uM:u.
`fl giants
`Ia ab-we
`B input: ax-porn,
`in cm. a frame,
`I] overrun. B isnm-ad.
`B input packets with xlrllzble cunditfinn dutncted
`1BB9 xmrz)u:L:2 uutpnt. 1fl§Y3ZI bytes, 8 under-rurls
`B nullu-jum-.-, 1 inlzax-Pam: »n.-nuts.
`IA rantarlra
`B uutnul: errnru,
`B output: husfav failu)-an,
`Ia uutpue buffers cirnppud nut
`set-ram in aw...
`line prutacnl in claim
`llnrdumxe in l|DE45'/M
`Internet nddl-ass is 1B.32.3.B 255.2-m.£1_n
`mu 15nIa bye:-.-, nu 31! )u:1t. DLY zflmm u!:I:i:. rely 255/255, land 1/255
`Ermapnulalzinu IIDLG,
`looyhnulr not set. Rucpnlive nut (10 Inc)
`Last input: nuuex-. nutput navel-. uutput hang Halter
`Last Eluarlny uf "nhtrw inf:erfm:e" auuniznra never
`/-ta, 2 drops;
`input queue Inns,
`:5 drop:
`autnuv. queue a
`5 minute ml
`pue nee B bits/23:, o patkatn/nan
`5 nmutc input rnlzu a hits/nae, u puckaeu/use
`:3 packets input, a bytes. a nu buffer-
`Racaiuad H broadcasts. E l‘Ill‘ltE. B niants
`Ia input‘. am-am, n cm,
`In Ermine. B uvavrun, B imwrnfl, B aluwt
`:3 packets nu¢—,m.c, o lrytts.
`El undnrrun:
`B collisions. 319 influx-Pace :-naets, B 1-astai-ts
`B nutput errors,
`_
`B uutput buffer hnluz-us, B n
`utput hu.fF\a1‘s szwaypad out
`I! I: under t;rannit:ion=
`DGD=doun
`an-inn in dual»D3K’lluIln
`lH‘R=I‘lnurI RIS=I‘loIm C'lS=(\nIm
`um» yr-ueuanl is dorm
`Ila:-duav: 1: N56457::
`
`Configuring a particular interface depends on the type of network
`protocol used by the network to which the interface port is con-
`nected. For example, an Ethernet port connected to an ll’ network
`will be configured for the routing of II’. An Ethernet port connected
`to an AppleTalk network will be configured for AppleTalk routing.
`Interface configuration is covered in Chapters 11, “Configuring IP
`Routing,” 12, "Routing Novel] IPX,” and 13, “Routing AppleTall-2.”
`SEE ALSO
`
`Crnmerting LAN zI)1dm'/'1/Ipon‘.r ta netzuar/c merlizr is zlixurrnerl on page I I9.
`
`7;.
`
`LAN ntetteces
`
`Cisco routers support several commonly used LAN networks. The
`most Common LAN router interfaces are Ethernet, Fast Ethernet,
`
`IBM Token Ring, and Fiber Distributed Data Interface (FDDI).
`
`All these LAN protocols mentioned embrace the same Data Link
`layer physical addressing system (the MAC hardware address on a
`NIC or the MAC hardware address found on the controller of the
`
`router interface). These addresses are unique for each device.
`
`109
`
`
`
`110
`
`
`
`PART II Router Design and Basic Configuration
`
`CHAPTER 6 Understanding Router interfaces
`
`networks use token passing as the network access strategy.
`Routers used on Token Ring networks must contain a special
`Token Ring interface for connection to the network. Parameters
`such as the speed of the ring (either 4Mbps or 16Mbps) must be
`set on the router’s Token Ring interface so the throughput speed
`matches that of the Token Ring network.
`
`FDDI is a token passing network that uses two redundant rings
`(passing tokens in opposite directions) as a fault tolerance
`method (fimlt tulemizce is keeping the network up and running
`when one of the rings breaks down). FDDI, which is often
`employed as a fiber—optic backbone for larger networks or
`municipal area networks (MANS), can provide network through-
`put of up to 100Mbps. Routers used on FDDI networks must
`have an FDDI interface.
`
`All the LAN protocols require a matching interface on the router
`that serves them. For example, a Tbken Ring network can only be
`attached to a router with the appropriate Token Ring interface.
`Specifications for some of the routers built by Cisco are discussed in
`Appendix C, “Selected Cisco Router Specifications.” You can also
`View the various specifications of Cisco routers on Cisco’s Vi/eb site
`at www. cisco. com. It is obviously important that when you plan your
`internetwork, you purchase a router that will provide you with all the
`necessary interfaces that your various LAN connections will require.
`Figure 6.5 shows the diagram of a network where several different
`LAN architectures have been connected using routers (the diagram
`is actually based on the network map of a real company’s internet—
`work).
`SEE ALSO
`ll/LAC (I(:'d7'e.\'.ver /We :1/'.rcI1.\'.red on page 41.
`
`2+»
`
`2» Frr/'7m71'e infbzwmtion an LAN /W/Jitctrzzrerrzzc/1 as Etliemet or FDDI, rcc page 25.
`
`Serial Interfaces V
`
`Serial router i11£c7fi1ces provide a way to connect LANS using WAN
`technologies. WAN protocols move data across asynchronous and
`synchronous serial interfaces (on routers), which are connected via
`leased lines and other third—party connectivity technologies.
`
`111
`
`
`
`PART II
`
`Seriai Interfaces CHAPTER 6
`
`{ IBM Token Ring Network x
`
`" “T
`1....
`——.:
`IBM Compaiible
`
`4,
`I
`1:...
`——-'4
`IBM Compalib
`
`——r+,
`V
`..
`\
`. L-‘J
`'
`T .r_¢»—
`IBM Compafib e IBM Compatible
`
`Clsco 4000 Router
`
`Cisco 2500 Router
`
`Cisco 2500 Router
`
`Ethernet Networks
`
`Some of the commonly used WAN Data Link layer technologies are
`High Level Data Link Control (HDLC), X.25 , Frame Relay,
`Integrated Services Digital Network (ISDN), and Point to Point
`Protocol (PPP). All the WAN protocols discussed are configured on
`particular router interfaces (such as a serial interface or an ISDN
`interface) when the router is in the configuration mode. The actual
`conunand sets and the ins and outs of configuring WAN protocols
`on a Cisco router are discussed in Chapter 15, “Configuring WAN
`Protocols.”
`
`n HDLC is a Data Link layer protocol that provides the encapsu-
`lation of data transferred across synchronous data links. This
`means that a device such as a DCE (Data Communication
`Equipment) provides a connection to the network and provides a
`clocking signal that synchronizes the transfer of data between
`
`112
`
`
`
`PART I| Router Design and Basic Configuration
`
`CHAPTER 6 Understanding Router Interfaces
`
`the two ends of the serial link. Serial ports on a router are con-
`nected to a modem or other CSU/DSU device via special cables
`such as 21 V35 cable. HDLC is the default VVAN protocol for
`Cisco routers. Cisco’s HDLC implementation is proprietary,
`however, and will not communicate with other vendor’s HDLC
`(this is why trying to mix routers from different vendors such as
`Cisco and 3Com can be a real nightmare). HDLC is considered
`a point—to—point protocol and provides :1 direct connection
`between sending and receiving devices (such as two routers).
`Point to Point Protocol (PPP) is another Data Link layer point‘
`to~point protocol supported by Cisco routers. It isn't proprietary,
`so it can be used to connect Cisco routers to internetworlcing
`devices from other vendors. PPP actually operates in both syn»
`chronous and asynchronous modes (meaning it can provide
`either encapsulation type). A flag (which is actually several bits
`inserted into the data stream) is used to signify the beginning or
`end of a frame or datagram of information flowing across the
`PPP connection. PPP can be used for connecting IP, AppleTalk,
`and IPX networks over WAN connections.
`
`PPP is configured on the serial port of the router that provides
`the connection to a leased line or some other WAN connection.
`You may already be familiar with PPP because it is the protocol
`used to connect workstations to Internet service providers over
`
`' analog phone lines via a modem.
`X.25 is a packet-switching protocol for use over public switched
`telephone networks. Data is passed along the switched network
`using virtual circuits (such as permanent virtual circuits). X25 is
`a slow protocol when compared to newer WAN technologies
`like Frame Relay because it provides a great deal of error checl(—
`ing (which was a must when X25 was first implemented several
`years ago over fairly low~grade telephone lines). X.25 is typically
`implemented between a DTE device and a DCE device. The
`DTE is typically a Cisco router, and the DCE is the X25 switch
`owned by the public switched network. Figure 6.6 shows how
`.
`two routers would be connected across an X.25 serial connec—
`tiO11.
`
`.
`
`’
`1
`
`V
`2
`f9|Y'0'Fl‘5T?‘Tt,i“'lFl
`?
`to‘ make sure thatthe data _
`.. is mmb|été|We..ceiV.ed by ii»
`the destination interface; V
`
`113
`
`
`
`PART II
`
`Serial Interfaces CHAPTER 6
`
`A
`
`such asX.25.
`
`;;
`
`'
`
`:1 Frame Relay is a packet switching Data Link layer protocol that
`was originally developed for use over ISDN connections. It has
`now replaced X.25 as the protocol of choice over switched net-
`works, and it uses Virtual circuits to define a route between two
`
`devices (such as two routers) communicating over the WAN. In
`:1 Frame Relay connection, 21 DTE such as a router is attached to
`a DCE such as a CSU/DSU(i11ost CSU/DSUS can be con-
`nected to the router using a V3 5 serial cable). Or the router can
`be connected directly to the phone cori1pany’s switching equip~
`merit. Frarne Relay[nd]based WANS looked similar to the X.25
`packet switching network depicted in Figure 6.6.
`_
`_
`1
`_
`,
`.
`lritegrated Services Digital Network (ISDN) uses digital tech-
`nology to move data, voice, and video over existing phone lines.
`It is an asynchronous WAN protocol. ISDN requires that the
`
`.
`._
`‘‘ typically comected to
`.~
`|sDN"p'o'mhgn3 prmgdgd,’ _
`5‘/3' 3P9“lfi° mmél
`’
`1 mum ‘A
`'
`‘
`'
`i
`
`ii
`
`114
`
`
`
`PART II Router Design and Basic Configuration
`
`CHAPTER 5 Understanding Router Interfaces
`
`network be connected to the phone line using terminal equip—
`ment that is commonly referred to as an ISDN modern.
`However, Cisco routers can be purchased that have a BRI inter-
`face (BRI stands for Basic Rate Interface) included on the router.
`The BRI interface is then connected directly to the phone lines.
`In cases where your router doesn’t have the BRI port, you will
`have to connect one of the existing serial ports to an ISDN
`modem (or buy a new router).
`SEE ALSO
`
`TV/lNp1‘aruvuls and bow 2‘/my 7:20/'lc /l7'L' divrllrred in grmter detail on page 53.
`
`tegieai interfaces
`Before we conclude our discussion of router interfaces, we must take
`at look at logical interfaces. A logical i7ZtE7ffl€€
`is a software—only inter-
`face and is created using the router’s IOS. Cisco’s lOS is explored in
`Chapter 9, “Working with the Cisco IOS.”
`
`Logical interfaces don’t exist as actual hardware interfaces on the
`router. You can think of logical interfaces as virtrml i17.Z'ETfiIC65‘ that
`have been created with a series of router software commands.
`
`These virtual interfaces can be viewed by devices on the network as
`real interfaces, just as a hardware interface such as a serial port is a
`real interface. You can configure different types of logical interfaces
`on a router including Loopback interfaces, Null interfaces, and
`Tunnel interfaces.
`
`toepbaek interfaces
`
`A Looplmc/e interfrzce is a software—only interface that emulates an
`actual physical interface on the router. Loopbacks are typically con-
`figured on a high—end router that serves as the core router between
`two corporate internetworks or between a corporate network and the
`Internet. Routers serving as core routers will be configured with an
`exterior gateway protocol such as Border Gateway Protocol that
`routes the packets between the two separate internetworks.
`
`Inteinetwork; "
`if
`V
`
`i
`
`115
`
`
`
`p
`
`A PART]!
`
`Logical Interfaces cl-|AP'l‘vE[‘{6
`
`Because the router serves as such an important link between»inter-
`networks, you don’t want it dumping data packets if a particular
`physical interface goes down on the router. So the Loopbackvirtual
`interface is created and configured as the termination address for the
`Border Gateway Protocol (BGP) sessions. In this way the traffic is
`processed locally on the router, which assures you that the packets
`get to their final destination.
`‘
`
`p
`
`Null Interfaces
`
`Another logical interface is the Null inteifrzce. It is set up on a router '-
`using the appropriate router commands and serves as a brick wall -
`that can be used to keep out certain traffic. For example, if you_don‘t
`want traffic from a particular network to move through a_ particular v
`router (but move through the internetwork by other routes) you can ‘
`configure the Null interface so that it receives and durnps any pack?
`ets that the network sends to the router. Normally Access lists.(dis~ i
`cussed in Chapter 14, “Filtering Router Traffic with Access Listsf’)
`are used to filter traffic on an internetwork a11d define valid routes
`for certain networks. The Null interface is pretty much a sledgeham— '
`met approach to a process that is normally handled with jeweler’s
`V
`‘
`tools.
`
`-
`
`Tunnel Interfaces
`‘used, to
`is another logical interface that can
`A Yimnel i7lt€7_'fflC€
`move packets of a particular type over a connection that doesn’t ty‘pi— V
`cally support these types of packets. For example, a Tunnel interface <
`can be set up on each of two routers that are responsible for routing _
`AppleTalk packets from their LANS. These two routers are con-_
`nected by a serial connection (see Figure 6.7). The Tunnel interface
`can he configured to route IP. And although AppleTalk would not be
`typically routed over anIP interface, the AppleTalk packets are ..
`encapsulated (stuffed in a generic envelope) and then moved across
`the Tunnel as if they were TP packets. Cisco routers provide the
`Generic Route Encapsulation Protocol (GRE), which handles the N
`encapsulation of packets moved over a Tunnel interface.
`’
`i
`
`116
`
`
`
`PART II Router Design and Bas'c Configuration
`
`CHAPTER 5 Understanding Hou er interfaces
`
`AppieTalk Network
`
`Macintosh
`
`Macintosh
`
`IP Network
`
`/
`
`Tunnel
`iP Serial Connection
`
`IP Network
`
`Macintosh
`Maclniosh Macintos
`App|eTaik Network
`
`117
`
`
`
`118
`
`
`
`119
`
`
`
`PART II
`
`Cisco Router Design CHAPTER "I
`
`(or routers) with the appropriate connections to fill your internet— .
`working requirements. (Many of the hi gher—end routers allow you to
`customize the type and number of interfaces found on the router.)
`
`fisee Roaster " esigu
`
`Cisco routers must be able to build routing tables, execute com-
`mands, and route packets across network interfaces using routing
`protocols. This means that the router must have processing power,
`some sort of storage capacity, and available random access memory.
`Appropriate software such as an operating system that can be used to
`configure routed and routing protocols is also necessary (and is dis-
`cussed in Chapter 9, “Working with the Cisco IOS”).
`
`Router CPUS
`
`Routers aren’t unlike PCs in that they contain a microprocessor. And
`just like PCs, different Cisco router models come with different
`processors. For example, the Cisco 2505 Router (which is the router
`that you will see in the various figures throughout this book) con-
`tains a 201\/IHz Motorola 68EC030 processor. A higher—end router
`like the Cisco 7010 Router contains a ZSMHZ Motorola MC68040 .
`
`CPU. (Many of the lower—end routers use some of the same
`Motorola processors that are used in a variety of Apple Macintosh
`computers. Some of the very high—end routers use Rise processors
`that you wou.ld typically find on miniframc computers or very high-
`end sewers.)
`SEE ALSO
`
`3:»
`
`lvbr more informm‘ian on rpzmfir Cim nmrers, rec page 33 7.
`
`Router Memory Components
`
`As already mentioned, routers not only need processing power, they
`also need a place to store configuration information, a place to boot
`the router operating system (IOS), and memory that can be used to
`hold dynamic information as the router does its job of moving pack-
`ets on the internetwork. Cisco routers actually contain different
`types of memory components that provide the storage and dynamic
`
`i
`__
`A ““°"”"°"‘ 3”“ 39”“
`tidns for”so‘me‘Of the Cisco "
`:_
`routers availahleg
`
`120
`
`
`
`PART]! Huuter Design and Basic Configuration
`
`CHAPTER 7 Setting Up a New Router
`
`caching required. The following list provides information on the dif-
`ferent memory components found in a Cisco router:
`
`in RONI—Contains the Power—on Self—Test (POST) and the boot-
`strap program for the router. The ROM chips also contain
`either a subset or the complete router IOS (for example, the
`ROl\/I on the 2505 router only contains a subset of the IOS,
`whereas the 7000 series contains the full IOS). Because the IOS
`is available on the ROAL you can recover irom major disasters
`such as the wiping out of your Flash RAM. The ROM chips on
`Cisco routers are removable and can be upgraded or replaced.
`
`NVRAM (nonvolatile RAM)~—Stores the startup configuration
`file for the router. NVRAM can be erased, and you can copy the
`running configuration on the router to NVRAM. The great
`thing about NVRAM is that it retains the information that it
`holds even if the router is powered down (which is extremely
`useful considering you won’t Want to have to reconfigure the
`router every time after the power goes down).
`
`Flash RAIVI——-Flash is a special kind of ROM that you can actu-
`ally erase and reprogram. Flash is used to store the Cisco IOS
`that runs on your router. You can also store alternative versions
`of the Cisco IOS on the Flash (such as an upgrade of your cur-
`rent IOS), which makes it very easy for you to upgrade the
`router, Flash RAM actually comes in the form of SINIMS
`(Single—Inline Memory Modules) and depending on the router
`you have, additional Flash RAM may be installed.
`
`RAM—Si111i.lar to the dynamic memory you use on your PC,
`RAM provides the temporary storage of information (packets are
`held in RAM when their addressing information is examined by
`the router) and holds information such as the current roufing
`table. RAM also holds the currently running router configura-
`tion (changes that you make to the configuration are kept in
`RAM until you save them to NVRAM).
`
`These various memory components all play an important role in
`What happens when you boot the router. The various possibilities
`revolving around the router system startup and where the router
`finds its [OS and start—up configuration files are discussed in the next
`chapter.
`
`121
`
`
`
`PART ll
`
`Connecting the Console c|=-iA‘PTER‘l
`
`S E E ALS 0
`
`' The role r/mt the different memory tflu-J play in the 7'a71terI1aot up sequerm: are disclosed in
`rbe wzexr rlmpm; beg/'11m'ng on page 126.
`SEE ALSO
`
`7779 Cisco Router inteifrrtex are mmr/yer impm-tmzt IJm'(I11)m'e campanelzr aftbe mute): T}J€_)’1?7‘E
`dixrzrxxcti in C11/Ipter 6, n‘/niing 0/1 page 99.
`
`flenneetiarg the Qeesee
`VVith an overview of the internal components of the router and the
`router interfaces
`the previous chapter) taken care of, it’s now time
`to walk through the steps of getting a new router out of its box and
`connecting it to the LANS that it will service (either by direct Con-
`nection using a LAN port such as an Ethernet port or by connecting
`LAN5 using WAN connections). Configuring the router is discussed
`in Chapter 8, “Basic Router Configuration,” with additional IOS
`configuration commands discussed in Chapters 9, 11, 12, 13, and 15.
`
`Before you attempt to conne