Unicast IP Routing
Unicast IP Routing |
|
The normal operation of a RIP for IP router consists of an initialization process, during which the router learns the routes of the internetwork from neighboring routers; an ongoing periodic advertisement process; and the proper advertisement of unreachable routes when the router is brought down through an administrative action.
Upon startup, the RIP for IP router announces its locally attached networks on all of its interfaces. The neighboring RIP routers process the RIP announcement and add the new network or networks to their routing tables as appropriate.
The initializing RIP router also sends a General RIP Request on all locally attached networks. The General RIP Request is a special RIP message requesting all routes. The neighboring RIP routers receive the General RIP Request and send a unicast reply to the requesting router. The replies are used to build the initializing RIP router's routing table.
By default, every 30 seconds the RIP router announces it routes on all of its interfaces. The exact nature of the routing announcement depends on whether the RIP router is configured for split horizon or split horizon with poison reverse. The RIP router is also always listening for RIP announcements from neighboring routers in order to add or update the routes in its own routing table.
If a RIP for IP router is downed properly through an administrative action, it sends a triggered update on all locally attached networks. The triggered update announces the networks available through the router with a hop count of 16 (unreachable). This topology change is propagated by neighboring RIP routers throughout the IP internetwork using triggered updates.
As dynamic routers, RIP for IP routers also react to changes in the internetwork topology from downed links or downed routers.
If a link goes down corresponding to one of the router's interfaces and this failure is detected by the interface hardware and indicated to the router, this change is sent out as a triggered update.
If a router goes down due to a power outage or other hardware or software failure, it does not have the ability to inform neighboring routers that the networks available through it have become unavailable. To prevent the lingering existence of unreachable networks in routing tables, each route learned by RIP for IP has a maximum lifetime of 3 minutes (by default). If the entry is not refreshed by the receipt of another announcement in 3 minutes, the entry's hop count is changed to 16 and it is eventually removed from the routing table.
Therefore, if a RIP for IP router goes down, it takes up to 3 minutes for the neighboring routers to mark the routes learned from the downed router as unreachable. Only then do they propagate the topology change throughout the internetwork using triggered updates.