Link-State Convergence

Link-State Convergence
Using Figure 1.2 as a reference, let’s now take a look at the convergence cycle used in link-state
routing protocols within a single area:
1. Router D detects the link failure between Routers D and F. The route entry for that link and
any dependent links are removed from Router D. A link-state advertisement (LSA) for OSPF,
or a link-state PDU (LSP) for IS-IS, is sent out all eligible OSPF or IS-IS interfaces on Router D.
2. Routers C and E receive the LSA or LSP and forward it out to all eligible interfaces, which
are normally all active interfaces except for the interface where the LSA was received, unless
Router C is the OSPF-designated router of the Ethernet network it shares with Router D
and Router D is not the backup designated router. If that’s the case, then Router C will
flood the LSA back out that interface also, as part of its duties as a designated router.
3. All routers wait five seconds, by default, and then run the shortest path first (SPF) algorithm.
After the algorithm is run, Router D adds the route through Router E, and Routers C, B, and
A update the metric in their routing table to that route.
Verifying and Testing a Route 29
4. After what could be another 30 seconds, Router F sends out an LSA or LSP to all eligible
OSPF or IS-IS interfaces after timing out the link to Router D. All routers wait five seconds
after receipt of the advertisement and then run the SPF algorithm, and all routers now know
that the route to the Ethernet segment off of Router F is through Router E.
Router A convergence time is the time of detection, plus the LSA forwarding time, plus five
seconds. This is about six seconds. However, if Router F’s time to converge is considered, then
the time can be about 36 seconds.
RFC 2328 is suggested reading for those interested in learning just about all they
ever wanted to know about how OSPF operates at the nuts-and-bolts level.