11.1.8 Notes

(1) Notes on Router ID, Network Addresses

The OSPF protocol uses a router ID to identify routers when constructing the network topology.

However, the protocol cannot construct an accurate topology when the network design contains either of the following irregularities:

• The same router ID is set for more than one router in a domain

• The same network address is assigned to different networks

If such an irregularity exists, the OSPF protocol cannot select paths accurately because the network design is based on inaccurate topology. We recommend using the following method to determine router IDs:

Determining router IDs

As the router ID of each router, choose one of the IP addresses assigned to an interface running OSPF on that router. The router ID can be any 32-bit numerical value, and this method can prevent a situation in which errors introduced during the design of the OSPF network cause router IDs to be duplicated.

When one router connects to more than one OSPF domain, using the same router ID in every domain does not present a problem.

(2) Notes on Route Redistribution and Learn Filters

The OSPF protocol advertises every LSA it learns from neighboring routers to its other neighbors. You cannot use redistribution filtering to prevent the advertisement of routes learned by OSPF within the same domain. Also, when using the route summarization functionality (the ip summary-address command) to summarize OSPF routes, you cannot prevent LSA advertisement within the same domain even if you configure route filtering to exclude summarization source routes from advertisement.

You can use the distribute-list in command to suppress learning of external AS routes that match the filter conditions. However, you cannot prevent OSPFv3 from learning and advertising LSAs. Therefore, OSPF will also advertise the routes it has not itself learned.

(3) Notes on using the multi-backbone facility