Routing Overview
Introduction
Load balancing and route summarisation work differently with different routing protocols, therefore it is important to understand convergence times, overhead and scaleability features. This lecture will also examine routing, default routes, floating static routes, convergence and route calculation.
This lecture will cover
Routing fundamentals
Static Routing
Configuring Dynamic Routing
Distance Vector Routing Protocols
Link State Routing
Hybrid Routing Protocol EIGRP
Default Routing Overview
Configuring Static Routes
Default Route Caveats
Configuring Floating Static Routes
Convergence
Route Calculation Fundamentals
Initiation of Routing Updates
Routing Metrics
Migration Issues
Configuring Default Routing with RIP and IGRP
Summary

Routing Fundamentals
Issuing the command

show ip route

makes a router display the contents of its routing table.

There are two types of routing - Dynamic Routing and Static Routing
Dynamic Routing
 Static Routing
Increased processor overhead Low processor overhead
Low maintenance  High maintenance
Increased network BW use - updates  No BW utilisation
Routers learn topology Predictable - packets go where specified
Learn new paths Secure - no false updates

No adaptability if a route fails

Static Routing
Static routes are useful for networks if there is no routing protocol in use. they have an Administrative Distance of 1 and a Cost of 0.
Can be configured as follows:

ip route [destination] [destination mask] [next hop address|interface]  [tag tag] [permanent]

The tag command allows for controlling redistribution using route maps.
The permanent command keeps the route in the routing table even if the interface shuts down.

Directly connected static routes are advertised in routing updates whereas those with next hop are not.

Static routing is unsuitable for large networks owing to the high administrative overhead owing to the multitude of routers and paths. The static routes would need to be configured on all routers.
Configuring Dynamic Routing
There are Two types of dynamic routing available for TCP/ IP, Interior Gateway Protocols e.g. OSPF, RIP, EIGRP, IGRP and Exterior Gateway Protocols e.g. EGP3, BGP4.

Legacy support exists for Novell IPX and APPLETALK
Novell (IPX) RIP - Distance Vector
Netware link Services Protocol NLSP - Link State
APPLETALK Routing Maintenance Protocol, RTMP - Distance Vector

The routing table contains much information concerning destination that the router knows about - see figure below.


An asterisk (*) next to a route in a routing table indicates the candidate default route.

Different routing protocols calculate the metric differently e.g. EIGRP uses BW and Reliability.

It is not possible to used dissimilar metrics when routes learned from different protocols appear in the same table to evaluate routes. In this case, Administrative Distance is used as the deciding metric. A lower value signifies a more trustworthy route or protocol that has supplied the route. The table below illustrates these values.

 

Route Source

Default Distance Values

Connected interface

0

Static route*

1

Enhanced Interior Gateway Routing Protocol (EIGRP) summary route

5

External Border Gateway Protocol (BGP)

20

Internal EIGRP

90

IGRP

100

OSPF

110

Intermediate System-to-Intermediate System (IS-IS)

115

Routing Information Protocol (RIP)

120

Exterior Gateway Protocol (EGP)

140

On Demand Routing (ODR)

160

External EIGRP

170

Internal BGP

200

Unknown**

255

Distance Vector Routing Protocols
Distance Vector routing protocols are based on the Bellman-Ford algorithm. They send a copy of the entire routing table at regular intervals as broadcasts on all interfaces.

upon receiving a table, the router compares it against its current table. New routes or those with better metrics are copied in and that router's next broadcast will include these updates. The router adds itself to the received hop count.

Distance Vector routing protocols are easier to configure than Link State routing protocols. They tend to use less memory and processing power. RIPv1 has almost universal support, but converges slowly and has a 15 hop count max.
Link State Routing
These routing protocols have greater scaleability and faster convergence than distance vector routing protocols.


Firstly the Link State router builds a topological database of all link states of every router in its area.


Each router makes its own map of the network. It then runs the Shortest Path First (SPF) algorithm on the topological database to produce an SPF tree with itself as the root.

Whenever a change occurs in the network topology, a partial update is sent to all participating routers at once.

Link state requires much memory and processing power.
Link state requires complex administration
Hybrid Routing Protocol EIGRP
Key features of EIGRP
Default Routing Overview
A Default Route is also known as the Gateway of Last Resort. A default route is used to keep routing tables small. It saves requiring a routing entry for all destination networks. A default route is set up as follows:

ip route 0.0.0.0 0.0.0.0
ip default-network
Configuring Static Routes

ip route 0.0.0.0 0.0.0.0 [next hop ip | interface to exit]

0.0.0.0/0 is known as the gateway of last resort  OR the quad zero route

Static routes to quad zero are sent in RIP updates.

IOS 12.1 and above does not propagate the static default routes.

Use network 0.0.0.0 to achieve this.

Alternatively use the:

default information originate {always}

command to propagate this address. this can be used with RIP and OSPF. The optional always distributes the route whether it is up or not.

Example

router(config)#ip route 0.0.0.0 0.0.0.0 172.16.1.2
router(config)#router rip
router(config-router)#default information originate {always}
Default Routing with IGRP
IGRP does not recognise 0.0.0.0/0 and will not send it in updates. To send default information in IGRP, use:

ip default-network

This distributes any route to an IP network, not just the default route. For example:

ip default-network 112.213.2.0


B is the boundary router of the IGRP 24 network. It needs to send default traffic to its link partner, the ISP router.

The command:

ip default-network 207.21.20.0

can be used here. This saves configuring routers A and C manually with a default route. A and C will also propagate this if there are other IGRP neighbours.

In a complex topology, several candidate default routes may be created. Once propagated, routers can choose the most suitable default route by choosing the one with the lowest cost. Candidate default routes are denoted by an asterisk (*).

ip default-network is classful.
Default Route Caveats

The diagram above illustrates how IGRP with VLSM can fail at times.

Routers do not use default routes for addresses that are part of their local domain i.e. major network to which the router is connected i.e. 172.16.0.0 /16.

Using IGRP, a static default route can be configured but because IGRP is classful, it sees a ping from router X to 172.16.1.1 /30 as local (belonging to the major network 172.16.0.0) and does not know how to deal with this as it does not use the default route.

Solutions
IOS v11.3 and above have ip classless enabled so subnets will be searched for and if not identified are sent to the quad zero route.

Alternatively, set up an explicit route  on router X to the major network 172.16.0.0

ip route 172.16.0.0 255.255.0.0 172.16.3.1

this will get round the problem too.
Configuring Floating Static Routes
Static routing cannot adapt to topology changes. A network can be configured to be more adaptable with floating static routes.

Static routes have a default admin distance of 1
Learned routes via RIP have an admin distance of 120

Therefore configure a static route with a higher admin distance than 120. The router will not use the static route unless the learned route fails.

therefore set up static routes with the least desirable route having the highest admin distance:

ip route 10.0.0.0 255.0.0.0 1.1.1.1 130

The figure of 130 above is the admin distance we have given to our route.
Convergence
When converged, all routers agree on the network topology. During convergence, the network may have routing problems because some routers routing tables are incorrect and contradict the tables in other routers leading to dropped packets or routing loops.
Help in design by using static default routes for stub networks. this saves so many routers being involved in the routing process.
Route Calculation Fundamentals
 Some routing protocols only allow for the presence of one route to a destination in the routing table at any time.
Multiple Routes to a single destination vs. a single route
Single Route in table
 Multiple Routes in table
Less processor overhead
 Greater processor overhead

Equal and unequal cost load balancing can be used

Less vulnerable to routing loops

Better recovery from failed links

No improvement in convergence time but more stable when link fails during convergence process



Initiation of Routing Updates
These can be time driven or event driven
Time Driven
 Event Driven
RIP - every 30 seconds
 EIGRP - sends partial updates
IGRP - every 90 seconds
 OSPF - sends partial updates

Time driven (for RIP) presents a delay of at most 30 seconds before an update is sent. We have to wait for up to 30 seconds for each router in a network to send an update meaning that convergence can be several minutes here. Loops or loss of connectivity are drawbacks here.

Exclusively event-driven routing protocols can go long periods without communicating. there must be a mechanism such as the Hello protocol to make sure routers are not working with out-of-date information.

Most routing protocols use a combination of time and event driven updates. Cisco RIP sends triggered updates when a change is detected. IGRP also sends triggered updates when a change is detected.

Event driven protocols use timers too. OSPF routers have a Max Age parameter on routing information. When this is exceeded, a new update must be requested.
Routing Metrics
A metric is a measure of the desirability of a route. Some of the factors used here are hop-count, BW, delay, load, reliability and MTU

BW and delay are static and only change if the network is redesigned.

Load and reliability are dynamic and calculated in real-time by the router.

The more  factors that are involved in a metric give a greater ability to adapt to network operation and react to specific needs.

IGRP default uses 2 static factors, BW and delay. 2 dynamic factors, load and reliability can also be set up in IGRP. When these are used, a link that becomes overloaded or unreliable can be bypassed in favour of another link that has a better metric i.e. a lower load or more reliable etc.

Migration Issues
It is possible to leave RIP running while setting up another routing protocol e.g. EIGRP.

Define an AS number and include EIGRP as the routing protocol.
Advertise it on networks
When using TELNET, use the terminal monitor command to see the output of debug ip routing

It is possible to clear the routing table with
clear ip route *

When EIGRP setup is complete, use no router rip to remove RIP.
Turn off debugging using

no debug all OR undebug all

Load Balancing Issues
IGRP requires that BW is specified to allow metric calculations, e.g.

int s0/0
bandwidth 256

Unequal cost load balancing is set up with the variance command.

router igrp 24
variance 5

This allows 5 times the lowest metric to be accepted

show ip protocols will display timers for routing protocols.

In a small network, adjust IGRP times DOWN to improve convergence.
Timer event
 Fast IGRP
 IGRP
Seconds between updates
 15s
 90s
Seconds for route expiration
 45s
 270s
Seconds for holddown
 0s
 280s
Seconds before route is flushed from routing table
 60s
 630s

Fast IGRP is set up using

router(config-router)#timers basic 15 45 0 60
router(config-router)#no metric holddown    -   this immediately accepts routes to networks that have been removed from the routing table
router(config-router)#metric maximum-hops 10  -  sets up the maximum diameter of the network

debug ip routing will report routing table changes to the console.
Configuring Default Routing with RIP and IGRP
A loopback interface simulates external connectivity on an interface. Good for testing purposes.

int loopback0
ip address 172.16.1.1 255.255.0.0

Loopback address can now be used as a destination address in its own right.

ip route 0.0.0.0 0.0.0.0 loopback0 sets up a static route to loopback0.

For RIP

router rip
default information originate

For IGRP, we need to use ip default network instead top propagate default information.

This makes IGRP flag the network as a candidate default network. The flag stays with the route as it is propagated between partners.

90 seconds may need to pass before an update, so use clear ip route * on all routers to force updates.

Summary

Routing protocols have two distinct types whether they are used within an AS or between ASs. These are IGP or EGP.

There are two types of routing protocol, distance Vector and Link State
Distance Vector
 Link State
RTMP
 OSPF
EIGRP - hybrid
 EIGRP - hybrid
IGRP
 ISIS
IPX RIP
 NLSP
RIPv1
RIPv2

Default and static routes can be useful to reduce routing table size, but are inflexible to change.

Floating static routes can be used in case of link failure and must be set up with a high administrative distance.

Routing updates can be configured manually - do this with care!