3.3 Routing Information Sources
In Figure 3-3, which shows three routers connecting four networks, router R1 would need additional information to be able to route to network 126.96.36.199. A line in R1's routing table would say that the network is accessible via router R2. How does this piece of information get into the routing table?
Routing information can get into the routing table in only two ways. As in any computer, information in a router can be either entered manually or gathered automatically. The administrator can investigate the network, sketch the topology, formalize the informationthat is, represent it in the form "For router R1, network N2 is accessible via router R3"and feed it to every router in the network. This is what happened at the beginning of the network world: Administrators used to watch the networks and edit the routing tables manually whenever the topology changed. This type of routing information, provided and entered into routers by administrators, is called static. Static routes reflect the administrators' knowledge of the topology and the policy they want to apply to the traffic going to a certain destination.
The drawbacks of static routing are obvious. Static routing generally cannot adapt to changes in topology or to load and error rates of channels. To have this adaptation, administrators would need to constantly keep an eye on the network and change the routing tables in real time. It may not seem difficult for a network with five routers, but imagine what would happen if this technique were used to manage a network consisting of hundreds of routers and channels connecting them. It would be impossible. This is the reason dynamic routing protocolsautomatic distributed network discovery and route calculation algorithmswere proposed, designed, and implemented.
The function of dynamic routing protocols is to exchange information about the networks routers know how to reach. Routing protocols provide routers with real-time information about where the networks are, which path is better for accessing a given network, which parallel paths can be used to balance the load among multiple channels, and so on. If routers have this information, they can maintain adequate routing tables.