RP in multicast

Unanswered Question
Giuseppe Larosa Sat, 02/21/2009 - 01:23

Hello Ali,

the RP rendez-vous point acts as a reference point to build a shared dsitribution tree.

In PIM sparse mode and other sparse mode multicast protocols the replication tree is rooted at the RP.

This together with explicit Joins allows for reduced state information on routers:

instead of building one tree for each (source, group) pair only one tree for each Group is built.

Traffic from sources is tunneled to the RP.

Then actually source based trees are built rooted at PIM routers connected to sources but the work done is re-used.

different nodes can be RP for different multicast addresses

Auto-RP: is the oldest dynamic way to map candidate RPs to groups and to propagate this info

pim v1.0

one router acts as mapping agent receives messages from RP candidates makes some choices and advertise a table of RP to groups mappings.

to work it requires

ip pim sparse-dense-mode

or ip pim autorp-listener

Boostrap protocol: modern dynamic mapping (pim v.2) to be preferred over auto-RP uses standard PIM messages, redundancy for the BSR router is supported instead auto-rp allows only one single mapping agent

auto-RP or BSR are to be used in big networks where is not practical to manually configure RPs on all routers in the network

Hope to help



This Discussion