Routing protocols allow routers to dynamically learn the topology of the network. Should the topology of the network change, routing protocols update their internal topology table, which allows them to route around failure.

There are two types of routing protocol, distance vector and link state. Distance vector protocols, including RIP and IGRP, exchange their entire routing table periodically. To each route, they add their metric (for RIP, this is hop count) and insert it in the routing table. If updates fail to arrive before the router’s timer expires, it purges the route and looks for another path. These protocols are usually slow to converge. See Sample convergence times (single link failure).

Link-state protocols, such as OSPF, take a more holistic view of the network. They compute the entire topology of the network and insert the best path to a destination in the routing table. Link state protocols exchange their routing tables only once, when routers first establish a relationship. After that, they only send updates. They also send hello messages periodically to ensure that the other routers are still present. Link state protocols converge much more quickly than distance vector protocols, and thus are generally better suited to networks that require high availability.