Designing a stable and scalable OSPF topology requires thought and planning. In larger networks, OSPF can have an impact on the workload of the router. The number of routers in an area, the number of adjacencies and the stability of the connections should be taken under consideration.
Talking about adjacencies, when designing OSPF topologies, especially in an environment with multiple LAN segments, distributing DR/BDR in different routers for each segment can reduce the load of a router significantly.
When we are thinking about designing areas, technics like summarization, stubby areas, and filtering can reduce as well the OSPF generated traffic and workload in the router. Summarization and stubby areas can have the same result where only a default route is advertised inside an area. Another design consideration when designing the topology is the area size. A general recommendation is to keep area 0 small and separate areas based on geographic or department consideration. Even if the hardware can handle a big number of areas in the same device, meaning that the router is ABR for multiple areas, keeping the LSDB for many areas, can cause issues.
A key aspect when talking about OSPF stability is the convergence time, the time from the detection of an event until having again a stable network. Having a small area, the convergence time can be optimal. However, there are ways to improve the detection of a failure faster and “cheaper” as far as resources are concerned. Using BFD can improve dramatically the detection of a link failure and improve the convergence with minimum impact on the resources.
