Cisco Systems 3560 사용자 설명서

Mapping agents periodically multicast the contents of their Group-to-RP mapping caches. Thus, all 
routers and switches automatically discover which RP to use for the groups that they support. If a router 
or switch fails to receive RP-discovery messages and the Group-to-RP mapping information expires, it 
changes to a statically configured RP that was defined with the ip pim rp-address global configuration 
command. If no statically configured RP exists, the router or switch changes the group to dense-mode 
operation.

Multiple RPs serve different group ranges or serve as hot backups of each other. 

PIMv2 BSR is another method to distribute group-to-RP mapping information to all PIM routers and 
multilayer switches in the network. It eliminates the need to manually configure RP information in every 
router and switch in the network. However, instead of using IP multicast to distribute group-to-RP 
mapping information, BSR uses hop-by-hop flooding of special BSR messages to distribute the mapping 
information.

The BSR is elected from a set of candidate routers and switches in the domain that have been configured 
to function as BSRs. The election mechanism is similar to the root-bridge election mechanism used in 
bridged LANs. The BSR election is based on the BSR priority of the device contained in the BSR 
messages that are sent hop-by-hop through the network. Each BSR device examines the message and 
forwards out all interfaces only the message that has either a higher BSR priority than its BSR priority 
or the same BSR priority, but with a higher BSR IP address. Using this method, the BSR is elected.

The elected BSR sends BSR messages with a TTL of 1. Neighboring PIMv2 routers or multilayer 
switches receive the BSR message and multicast it out all other interfaces (except the one on which it 
was received) with a TTL of 1. In this way, BSR messages travel hop-by-hop throughout the PIM 
domain. Because BSR messages contain the IP address of the current BSR, the flooding mechanism 
enables candidate RPs to automatically learn which device is the elected BSR.

Candidate RPs send candidate RP advertisements showing the group range for which they are 
responsible to the BSR, which stores this information in its local candidate-RP cache. The BSR 
periodically advertises the contents of this cache in BSR messages to all other PIM devices in the 
domain. These messages travel hop-by-hop through the network to all routers and switches, which store 
the RP information in the BSR message in their local RP cache. The routers and switches select the same 
RP for a given group because they all use a common RP hashing algorithm.

With unicast routing, routers and multilayer switches forward traffic through the network along a single 
path from the source to the destination host whose IP address appears in the destination address field of 
the IP packet. Each router and switch along the way makes a unicast forwarding decision, using the 
destination IP address in the packet, by looking up the destination address in the unicast routing table 
and forwarding the packet through the specified interface to the next hop toward the destination.

With multicasting, the source is sending traffic to an arbitrary group of hosts represented by a multicast 
group address in the destination address field of the IP packet. To decide whether to forward or drop an 
incoming multicast packet, the router or multilayer switch uses a reverse path forwarding (RPF) check 
on the packet as follows and shown in 

:

The router or multilayer switch examines the source address of the arriving multicast packet to 
decide whether the packet arrived on an interface that is on the reverse path back to the source.

If the packet arrives on the interface leading back to the source, the RPF check is successful and the 
packet is forwarded to all interfaces in the outgoing interface list (which might not be all interfaces 
on the router).