A review about routing protocols

An Introduction

A routing protocol specifies how routers communicate with each other, distributing information that enables them to select routes between any two nodes on a computer network. Routing algorithms determine the specific choice of route. Each router has a prior knowledge only of networks attached to it directly. A routing protocol shares this information first among immediate neighbors, and then throughout the network. This way, routers gain knowledge of the topology of the network.

The specific characteristics of routing protocols include the manner in which they avoid routing loops, the manner in which they select preferred routes, using information about hop costs, the time they require to reach routing convergence, their scalability, and other factors.

Although there are many types of routing protocols, three major classes are in widespread use on IP networks:

• Interior gateway protocolstype 1, link-state routing protocols, such as OSPF and IS-IS


• Interior gateway protocolstype 2, distance-vector routing protocols, such as Routing Information Protocol, RIPv2, IGRP.


• Exterior gateway protocolsare routing protocols used on the Internet for exchanging routing information between Autonomous Systems, such as Border Gateway Protocol (BGP), Path Vector Routing Protocol. Exterior gateway protocols should not be confused with Exterior Gateway Protocol (EGP), an obsolete routing protocol.

 

 

A routing protocol uses software and routing algorithms to determine optimal network data transfer and communication paths between network nodes. Routing protocols facilitate router communication and overall network topology understanding.

• A routing protocol is also known as a routing policy.

Most Internet Protocol

(IP) networks use the following routing protocols:

• Routing Information Protocol (RIP) and Interior Gateway Routing Protocol (IGRP): These provide interior gateway routing through path or distance vector protocols.


• Open Shortest Path First (OSPF): This provides interior gateway routing through link-state routing protocols.


• Border Gateway Protocol (BGP) v4: This provides public Internet routing protocol through exterior gateway routing

• A routed protocol is a protocol by which data can be routed. Routed protocol are IP, app letalk, and IPX. In this kind of protocols we require an addressing scheme and sub netting. Addressing scheme will be used to determine the network to which a host belongs and to identifying that host on that particular network. All hosts on an internetwork are using the services of a routed protocol. That means routers, servers, and workstations to. The only two routed protocols that are in use today are IP and IPX but IPX is dropped from Cisco in exams and is not in use much these days. If you are studying routed protocols the best advice is to focus on IP routed protocol.

 

 

A routing protocol is different and is only used between routers. It makes possible for routers to build and maintain routing tables. There are three classes of routing protocols-

1) Distance vector,

2) Link state,

3) Hybrid

OSPF is one of two link state protocols, the other one is IS-IS. EIGRP is the only hybrid protocol but in normal literature you will see that EIGRP is distance vector routing protocol.

Configuring Routing Protocols

A routing protocol needs to know three pieces of information, at a minimum:

• Who am I (my identity)


• To whom to disseminate information


• What to disseminate

Most routing protocols use the concept of a router ID to identify a node. Different routing protocols answer the last two questions differently.

The way they answer these questions affects the network design and thereby configuration. For example, in a link-state protocol such as OSPF (see Open Shortest Path First (OSPF) Protocol) or IS-IS, complete local information (links and attached address prefixes) about a node is disseminated to every other node in the network. Since the state that a node has to keep grows rapidly in such a case, link-state protocols typically limit the number of nodes that communicate this way. They allow for bigger networks to be built by breaking up a network into a set of smaller subnetworks (which are called areas or levels), and by advertising summarized information about an area to other areas.

Besides the two critical pieces of information mentioned above, protocols have other parameters that can be configured. These are usually specific to each protocol.

Protocol Tuning

Most protocols provide certain tunable parameters that are specific to convergence during changes.

Wikipedia defines convergence as the “state of a set of routers that have the same topological information about the network in which they operate”. It is imperative that the routers in a network have the same topological state for the proper functioning of a network. Without this, traffic can be blackhead, and thus not reach its destination. It is normal for different routers to have differing topological states during changes, but this difference should vanish as the routers exchange information about the change and recomputed the forwarding paths. Different protocols converge at different speeds in the presence of changes.

Figure 1 Routing example

A key factor that governs how quickly a routing protocol converges is the time it takes to detect the change. For example, how quickly can a routing protocol be expected to act when there is a link failure. Routing protocols classify changes into two kinds: hard changes such as link failures, and soft changes such as a peer dying silently. They’re classified differently because protocols provide different mechanisms for dealing with these failures.

It is important to configure the protocols to be notified immediately on link changes. This is also true when a node goes down, causing all of its links to go down. Even if a link doesn’t fail, a routing peer can crash. This causes that router to usually delete the routes it has computed or worse, it makes that router impervious to changes in the network, causing it to go out of sync with the other routers in the network because it no longer shares the same topological information as its peers.

The most common way to detect a protocol peer dying is to detect the absence of a heartbeat. All routing protocols send a heartbeat (or “hello”) packet periodically. When a node does not see a consecutive set of these hello packets from a peer, it declares its peer dead and informs other routers in the network about this. The period of each heartbeat and the number of heartbeats that need to be missed before a peer is declared dead are two popular configurable parameters. If you configure these timers very low, the network can quickly descend into instability under stressful conditions when a router is not able to keep sending the heartbeats quickly as it is busy computing routing state; or the traffic is so much that the hellos get lost. Alternately, configuring this timer to very high values also causes blackholing of communication because it takes much longer to detect peer failures. Usually, the default values initialized within each protocol are good enough for most networks. Cumulus Networks recommends you do not adjust these settings.

 

Packets may pass through several networks on their way to destination. Each network carries a price tag, or a “metric”. The metric of a network may be constant (i.e. each network costs one hop) Service type-dependent (the cost of the network depends on what service the packet needs: e.g. throughput, delay etc.)