Saturn and Mars to get to the 10.0.0.0/8 network attached to the Earth router.

EIGRP has a lower administrative distance (90) than OSPF (110) and is therefore

preferred.

Metrics

In the previous example, two routing protocols run on the routers, but OSPF

and EIGRP chose two different paths to get to the Earth router. Each routing

protocol has its own algorithm to determine what it considers to be the best path

to a destination network. The main factor in deciding the best path is the routing

Introduction

A metric is the variable used in the algorithm when making routing decisions.

Each routing protocol uses a different type of metric. Table 10.2 illustrates the

different metrics used by routing protocols.

RIP Hop Count The number of hops, or routers, that a packet

has to pass through to reach a destination. The

route with the lowest hop count is preferred.

EIGRP Bandwidth, Delay Uses Bandwidth and Delay by default, but also

can factor Reliability, Load, and Maximum

Transmission Unit (MTU).

OSPF Cost Cost is defined as 108/bandwidth.

Metrics are not the only thing that distinguishes the routing protocols. Routing

protocols can be further classified into two categories:

Distance Vector Routing Protocols

Distance vector routing protocols include RIP and the now unsupported legacy

protocol, Interior Gateway Routing Protocol (IGRP). EIGRP is a hybrid that

contains many of the characteristics of a distance vector protocol. Characteristics

of distance vector routing protocols are as follows:

routing is sometimes called “routing by rumor.”)

Because distance vector routing protocols trust the next router without compiling

a topology map of all networks and routers, distance vector protocols run the

risk of creating loops in a network.

This is analogous of driving to a location without a map. Instead, you trust what

each sign tells you. Trusting the street signs might get you where you want to go,

but I’ve been in some cities where trusting what the signs say will lead you in loops.

The same is true with distance vector routing protocols. Simply trusting what the

next router tells it can potentially lead the packets to loop endlessly. These loops

could saturate a network and cause systems to crash. This, in turn, makes managers

very upset and means that you have to work late into the evening to fix it.

Luckily, distance vector protocols have some mechanisms built in to them to

prevent loops. These mechanisms are as follows:

Routers maintain a routing table which is stored in RAM. The routing table

lists every network the router has learned about and the number of hops, or

routers, it takes to go through to get to a destination network. For example, if a

packet sent from a router needs to go through two other routers to get to the

destination network, a hop count of two would be recorded. All distance vector

routing protocols maintain a record of hop count even if they do not use hop

count in their routing decisions.

Examine Figure 10.4. Through the use of a dynamic routing protocol, each

router will exchange information with the next router. Mars will learn of the

networks known by Saturn and Jupiter, and Mars will let Saturn and Jupiter

know of the networks that Mars knows about. The next table shows the networks

and associated hop counts for each router.

Distance vector routing protocols keep track of hop counts because if a route

exceeds a maximum hop count limit (determined differently by each routing

protocol), the network is considered unreachable. This prevents packets from

cycling endlessly across your networks. Table 10.4 shows the maximum hop

count for distance vector protocols.

Having a maximum hop count should be enough to prevent loops, but because

loops are so dangerous, other methods are used as well. The second method to

prevent routing loops is split horizon. The split horizon rule states that information

about a route should not be sent back in the direction in which it was learned.

Look back at Figure 10.4. The split horizon rule states that if Saturn tells Mars

about the 13.0.0.0/8 network, Mars should not advertise it back to Saturn. If it

did, Saturn would be confused and think that it could possibly use Mars to get

to the 13.0.0.0 should its interface to that network ever go down. This would

cause a packet to loop endlessly as the packet would go to Mars, which would in

turn send it back to Saturn. Split horizon resolves this issue by ensuring that the

Mars router never sends information about the 13.0.0.0 network back to the

Saturn router that it heard it from.

To make absolutely sure that no loops are created, route poisoning and poison

reverse are also implemented. With route poisoning, as soon as a network is

thought to be down, it is advertised out with a hop count that is one greater than

what is allowed. This would declare the route as being inaccessible. Poison

reverse does the same thing but in reverse. The router that hears about a down

network, violates split horizon, and sends back an update with the network being

unreachable. Figure 10.5 illustrates how this would look if the routers were running

RIP, where the maximum hop count is 15 and a hop count of 16 declares

the route inaccessible.

The next mechanism to prevent loops is holddown timers. When a router

receives information that a network is possibly down from a neighbor router, it

will not accept any new information from that router for a specified period of

time. This is to prevent regular update messages from reinstating a down route.

The default holddown timer for RIP is 180 seconds.

Finally, triggered updates are used to prevent loops by exchanging routing information

whenever there is a change. In other words, a change in the routing

topology will trigger routers to update each other. Without triggered update, a

router would have to wait for the next update interval to learn of a changed route.

During that period when a route is changed and when the next routine update is

sent out there is a potential of a loop. To lessen the risk of a loop during this waiting

period, routers will not wait for the update interval to send out the information

about a changed network but will instead send out the information

immediately. This way all routers can learn of the change as soon as possible.

If distance vector routing protocols are like trusting the highway signs when you

are on a road trip, link state routing protocols are like having the map in front

of you. With link state routing protocols such as OSPF, your router will know

all the networks and the various paths to the networks.

Extra! Extra! Read all about it! EIGRP solves the world’s problems. It’s the best

of both worlds! You get the best of link state and distance vector routing all built

in to one protocol!

Okay, so perhaps that’s a little more hype than necessary, but it is not that far

from the truth. EIGRP is a Cisco-proprietary protocol that combines characteristics

of link state and distance vector routing protocols. For example, like a link

state routing protocol, it sends out hello messages to discover its neighbors.

However, it does not have a built-in hierarchical design like OSPF, thus making

it more like a distance vector. The operations and configurations of EIGRP and

OSPF are not tested on in the ICND1 exam, but you will want to know the differences

between link state and distance vector protocols. You read more about

EIGRP later, but for now let’s start with a very simple protocol, RIP.

The Routing Information Protocol (RIP) uses the Bellman-Ford algorithm,

which simply counts the number of hops, or routers, to a destination network

and chooses the path that is the fewest number of hops. Any destination that is

more than 15 hops away is considered inaccessible.

RIP routers exchange information by broadcasting the entire routing table every

30 seconds out all interfaces with RIP enabled. RIP version 2 also sends out

updates every 30 seconds but sends out updates using the multicast address of

224.0.0.9 (can be configured to do unicast as well). In addition, version 2 provides

the following benefits not available in version 1:

. Routing authentication

. Classless routing

. Summarization

Configuring RIP is straightforward. The four steps to configuring a routing

protocol are as follows:

1. Enable the routing protocol.

2. Activate it on interfaces.

3. Advertise directly on networks.

4. Configure optional parameters.

mode by typing router rip. The next two steps, activating RIP on interfaces and

advertising networks, is done with a single command, the network command.

Moe(config)#router rip

Moe(config-router)#network 192.168.10.0

Moe(config-router)#network 192.168.20.0

Larry has three networks attached to his router. His configuration would be

Larry(config)#router rip

Larry(config-router)#network 192.168.20.0

Larry(config-router)#network 192.168.30.0

Larry(config-router)#network 192.168.40.0

Finally, we can’t forget Curly. Curly’s configuration would be

Curly(config)#router rip

Curly(config-router)#network 192.168.40.0

Curly(config-router)#network 192.168.50.0

When you enter your networks in your RIP configuration, RIP is activated on

the interfaces that are assigned those networks. All networks that you listed in

your configuration are then sent out all RIP-activated interfaces. Thus, the networks

that you entered on Curly’s router will be sent out to Larry. Larry will

take what he learned from Curly, add his own networks, and send them out to

Moe. Larry will also learn networks from Moe, add his own networks, and send

them out to Curly.

Remember to enter only your directly connected networks. Curly, for example,

should not enter 192.168.10.0/24 in his configuration because that network is

not directly connected to his router. Also, you should enter classful networks

only. A classful network is the major class A, B, or C network with the default

masks of /8, /16, or /24. This means that even if you are subnetting, you should

enter the major Class A, B, or C address. In Figure 10.7, our three friends have

new networks that are taken from a major Class A network. Even though multiple

networks are attached to them, enter only the major 10.0.0.0/8 network.

Thus, all three routers would have the same configuration:

Router(config)#router rip

Router(config-router)#network 10.0.0.0

Finally, you may enter some optional commands. The two optional commands

that you should be familiar with for the exam are as follows:

. version 2

. no auto-summary

Both commands are entered under the RIP routing process. The first command,

version 2, enables RIP version 2 on your router. RIP version 2 adds the benefits

network.) To disable automatic summarization, enter the no auto-summary command

under the routing process. Using Figure 10.7 again, the complete configuration

for Larry’s router, assuming that you wanted RIP version 2 with no

automatic summarization, is

Larry(config)#router rip

Larry(config-router)#network 10.0.0.0

Larry(config-router)#version 2

Larry(config-router)#no auto-summary

Note that even though we disabled automatic summarization, we still put the

default classful networks in our configuration. RIP is smart enough to go on the

interfaces and discover the individual subnetworks and their associated subnet

masks.

Now that RIP is configured, you should verify your configuration. There are

two commands that you can use to verify proper operation of RIP:

. show ip route

. show ip protocols

The first command displays your routing table. For the sake of simplicity, we’ll

go back to our original example of our three friends before they got creative and

started subnetting. Figure 10.8 shows the Larry, Curly, and Moe routers before

they subnetted. This time, the names of the interfaces have been included.

After executing the show ip route command on Larry’s router, you should see

the following:

Gateway of last resort is not set.

R 192.168.10.0 [120/1] via 192.168.20.1 00:00:08, Serial 0/0

R 192.168.50.0 [120/1] via 192.168.40.2 00:00:16, Serial 0/1

C 192.168.30.0 is directly connected, FastEthernet 0/0

C 192.168.20.0 is directly connected, Serial 0/0

C 192.168.40.0 is directly connected, Serial 0/1