RP Mapping Options
RPs can be learned by one of the following mechanisms.
Static Configuration
You can configure a static RP configuration that is very similar to static routes. A static configuration has the benefit of operating in PIM version 1 or version 2. When you configure the static RP, the RP address that you select for a particular group must be consistent across all routers in a multicast domain.
A static configuration is simple and convenient. However, if the statically defined RP router becomes unreachable, there is no automatic failover to another RP router. To remedy this problem, you can use anycast RP.
Anycast RP
Anycast means that multiple RP routers share the same unicast IP address. Anycast addresses are advertised by the routing protocols. Packets sent to the anycast address are sent to the nearest RP with this address. Anycast addressing is a generic concept and is used in PIM sparse mode to add load balancing and service reliability to RPs.
Having a single active RP per multicast group is much the same as having a single server providing any service. All traffic converges on this single point, although other servers are sitting idle, and convergence is slow when the resource fails. In multicast specifically, there might be closer RPs on the shared tree, so the use of a single RP is suboptimal.
When anycast RP is configured, the shared address is used in the RP-to-group mapping. This allows multicast groups to have multiple active RPs in a PIM domain. However, the RPs must use some protocol to synchronize the active source information so that the active RP for each group is known to all RPs.
There are two methods for RP active source synchronization in anycast RP, one using the Multicast Source Discovery Protocol (MSDP) and the other using PIM itself.
When MSDP is used with PIM sparse mode, anycast RP provides a faster failover rate than auto-RP or a bootstrap router. However, MSDP only works for IPv4. When PIM alone is used for anycast RP, the solution works for both IPv4 and IPv6.
For more information about configuring static RPs, see Configuring Static PIM RPs. For more information about configuring anycast RP, see Configuring PIM Auto-RP and Example: Configuring PIM Anycast with or without MSDP.
Auto-RP
You can configure a more dynamic way of assigning RPs in a multicast network by means of auto-RP. When you configure auto-RP for a router, the router learns the address of the RP in the network automatically and has the added advantage of operating in PIM version 1 and version 2.
Although auto-RP is a nonstandard (non-RFC-based) function that typically uses dense mode PIM to advertise control traffic, it provides an important failover advantage that simple static RP assignment does not. You can configure multiple routers as RP candidates. If the elected RP stops operating, one of the other preconfigured routers takes over the RP functions. This capability is controlled by the auto-RP mapping agent.
For more information, see Configuring PIM Auto-RP.
Bootstrap Router
To determine which router is the RP, all routers within a PIM sparse-mode domain collect bootstrap messages. A PIM sparse-mode domain is a group of routers that all share the same RP router. The domain's bootstrap router originates bootstrap messages, which are sent hop by hop within the domain. The routers use bootstrap messages to distribute RP information dynamically and to elect a bootstrap router when necessary.
For more information, see Configuring PIM Bootstrap Properties.
