Static RP and RP Redundancy

You can provide RP redundancy through static RPs. To ensure consistency of RP selection, implement the same static RP configuration on all PIM-SM routers in the network. In a mixed vendor network, ensure that the same RP selection criteria is used among all routers. For example, to select the active RP for each group address, the switch uses a hash algorithm defined in the PIM-SMv2 standard. If a router from another vendor selects the active RP based on the lowest IP address, then the inconsistency prevents stream delivery to certain routers in the network.

If a group address-to-RP discrepancy occurs among PIM-SM routers, network outages occur. Routers that are unaware of the true RP cannot join the shared tree and cannot receive the multicast stream.

Failure detection of the active RP is determined by the unicast routing table. As long as the RP is considered reachable from a unicast routing perspective, the local router assumes that the RP is fully functional and attempts to join the shared tree of that RP.

The following figure shows a hierarchical OSPF network where a receiver is in a totally stubby area. If RP B fails, PIM-SM router A does not switch over to RP C because the injected default route in the unicast routing table indicates that RP B is still reachable.

Click to expand in new window

Because failover is determined by unicast routing behavior, carefully consider the unicast routing design, as well as the IP address you select for the RP. Static RP failover performance depends on the convergence time of the unicast routing protocol. For quick convergence, ensure that you use a link state protocol, such as OSPF. For example, if you use RIP as the routing protocol, an RP failure can take minutes to detect. Depending on the application, this situation can be unacceptable.

Static RP failover time does not affect routers that have already switched over to the SPT; failover time only affects newly-joining routers.