The Protocol Independent Multicasting-Sparse Mode (PIM-SM) is a multicast routing protocol designed to operate efficiently across Wide Area Networks (WANs) with sparsely distributed groups. It helps geographically dispersed network nodes to conserve bandwidth and reduce traffic by simultaneously delivering a single stream of information to multiple locations. PIM-SM uses the IP multicast model of receiver-initiated membership, supporting both shared and shortest-path trees, and uses soft-state mechanisms to adapt to changing network conditions. It relies on a topology-gathering protocol to populate a multicast routing table with routes.

Following is a brief description of terms and concepts used to describe the PIM-SM protocol:

A Rendezvous Point (RP) router is configured as the root of a non-source-specific distribution tree for a multicast group. Join messages from receivers for a group are sent towards the RP. Data from senders is sent to the RP so that receivers can discover who the senders are, and receive traffic destined for the group.

The Multicast Routing Information Base (MRIB) is a multicast topology table derived from the unicast routing table. In PIM-SM, the MRIB decides where to send Join/Prune messages. It also provides routing metrics for destination addresses. These metrics are used when sending and processing Assert messages.

Reverse Path Forwarding (RPF) is an optimized form of flooding, in which the router accepts a packet from SourceA through Interface IF1, only if IF1 is the interface the router uses to reach SourceA. To determine if the interface is correct, it consults its unicast routing tables. The packet that arrives through interface IF1 is forwarded because the routing table lists this interface as the shortest path. The router's unicast routing table determines the shortest path for the multicast packets. Because a router accepts a packet from only one neighbor, it floods the packet only once, meaning that (assuming point-to-point links) each packet is transmitted over each link, once in each direction.

The Tree Information Base (TIB) is a collection of states at a PIM router storing the state of all multicast distribution trees at that router. The TIB is created by receiving Join/Prune messages, Assert messages, and IGMP information from local hosts.

Upstream indicates that traffic is going towards the root of the tree. The root of the tree might be either the Source or the RP.

Downstream indicates that traffic is going away from the root of the tree. The root of tree might be either the Source or the RP.

In Source-Based Trees, the forwarding paths are based on the shortest unicast path to the source. If the unicast routing metric used is hop counts, the branches of the multicast Source-Based Trees are minimum hop. If the metric used is delay, the branches are minimum delay. A corresponding multicast tree directly connects the source to all receivers for every multicast source. All traffic to the members of an associated group passes along the tree made for their source. Source-Based Trees have two entries with a list of outgoing interfaces -- the source address and the multicast group.

Shared trees, or RP trees (RPT), rely on a central router called the Rendezvous Point (RP) that receives all traffic from the sources, and forwards that traffic to the receivers. There is a single tree for each multicast group, regardless of the number of sources. Only the routers on the tree know about the group, and information is sent only to interested receivers. With an RP, receivers have a place to join, even if no source exists. The shared tree is unidirectional, and information flows only from the RP to the receivers. If a host other than the RP has to send data on the tree, the data must first be tunneled to the RP, then multicast to the members. This means that even if a receiver is also a source, it can only use the tree to receive packets from the RP, and not to send packets to the RP (unless the source is located between the RP and the receivers).

When a new multicast sender starts sending data packets, or a new receiver starts sending Join messages towards the RP for that multicast group, the sender needs to know the next-hop router towards the RP. The bootstrap router (BSR) provides group-to-RP mapping information to all the PIM routers in a domain, allowing them to map to the correct RP address.

PIM-SM is a soft-state protocol. The required steps to configure PIM-SM are the following:

All multicast group states are dynamically maintained as the result of IGMP Report/Leave and PIM Join/Prune messages.

This section provides the steps to configure the PIM-SM feature. Configuration steps and examples are used for two relevant scenarios.

The following figure displays the network topology used in these examples.

Enable IP multicast routing on all of the PIM routers inside the PIM domain:

Enable PIM-SM on all participating interfaces within each of routers inside the PIM domain on which you want to run PIM. In the following sample configuration, both eth1 and eth2 are enabled for PIM-SM on the router.

Every PIM multicast group needs to be associated with the IP address of a Rendezvous Point (RP), which is a router that resides in a multicast network domain. The address of the RP is used as the root of a group-specific distribution tree. All nodes in the domain that want to receive traffic sent to the group are aware of the address of the RP. For all senders to reach all receivers within a group, all routers in the domain must be able to map to the RP address configured for the group. There can be several RPs configured in a network deploying PIM-SM, each serving a different group.

You can statically configure a RP by specifying the RP address with in every router in the PIM domain. The use of statically configured RPs is ideal for small network environments or ones that do not require many RPs and/or require changing the assignment of the RPs often. Changing the assignment of an RP requires the re-configuration of the RP address in all of the routers in the PIM domain.

In static RP configurations, RP failover is not available.

When configuring the RP statically, do the following:

Using the topology depicted in Figure 3-1, Router_C is the RP, and all routers are statically configured with RP information. Host_1 and Host_2 join group 224.0.1.3 for all the sources. They send the IGMP membership report to Subnet 1. Two routers are attached to Subnet 1, Router_E and Router_F; both have default DR priority on eth1. Since Router_E has a higher IP address on interface eth1, it becomes the Designated Router, and is responsible for sending Join messages to the RP (Router_C).

Here is the sample configuration for Router_D:

Enter the commands listed in this section to confirm the previous configurations.

At Router_D, the show ip pim rp mapping command shows that 10.10.1.5 is the RP for all multicast groups 224.0.0.0/4, and is statically configured. All other routers will have a similar output:

Override RP cnt: 0At Router_D, use the show ip pim rp-hash command to display the selected RP for a specified group (224.0.1.3):

The show ip pim interface command displays the interface details for Router_E, and shows that Router_E is the Designated Router on Subnet 1.

The show ip pim mroute command displays the IP multicast routing table. In this table, the following fields are defined:

At Router_E, eth2 is the incoming interface of the (*, G) entry, and eth1 is on the outgoing interface list of the
(*, G) entry. This means that there is a group member through eth1, and the RP is reachable through eth2.

The 0 position on this 32-bit index is for eth1 (as illustrated in the interface display above). The j on the 0 index indicates that the Join has come from eth1.

Since Router_C is the RP, and the root of this multicast tree, the show ip pim mroute command on Router_C shows RPF nbr as 0.0.0.0 and RPF idx as none.

A static RP configuration works for a small, stable PIM network domain; however, it is not practical for a large and/or complex one. In such a network, if the RP fails or you have to change the assignment of the RP, you are required to reconfigure the static configurations on all PIM routers. Also, if you have several multicast groups mapped to several RPs, there are many repetitive configurations you are required to perform, which can be time consuming and laborious. Thus when it comes configuring RP in large and/or complex networking environments, configuring it dynamically is the best and most scalable method to use. Bootstrap router (BSR) configuration is one method of configuring the RP dynamically.

The BSR mechanism in a PIM domain uses the concept of a RP as a way for receivers to discover the sources that send to a particular multicast group. The BSR mechanism gives a way for a multicast router to learn the set of group-to-RP mappings required in order to function. The BSR’s function is to broadcast the RP set to all routers in the domain.

Some of the PIM routers within a PIM domain are configured as Candidate-RPs (C-RPs). A subset of the C-RPs is eventually used as the actual RPs for the domain. An RP configured with a lower value in the priority field has a higher priority.

Some of the PIM routers in the domain are configured to be Candidate-BSRs (C-BSR). One C-BSR is selected to be the BSR for the domain, and all PIM routers in the domain learn the result of this election through Bootstrap messages (BSM). The C-BSR with highest value in the priority field is elected to be the BSR. The C-RPs then report their candidacies to the elected BSR, which chooses a subset of the C-RPs, and distributes corresponding group-to-RP mappings to all the routers in the domain using Bootstrap messages.

This section provides 2 examples to illustrate the BSR configuration for configuring RP dynamically.

For this example, refer to Figure 1 for the topology.

To dynamically configure the RP, Router_C on eth1 and Router_D on eth1 are configured as a Candidate RP using the ip pim rp-candidate command. Router_D on eth1 is also configured as the Candidate BSR. Since no other router has been configured as the candidate BSR, Router_D becomes the BSR router and is responsible for sending group-to-RP-mapping information to all other routers in this PIM domain.

The highest priority router (configured with lowest priority value) is chosen as the RP. If two or more routers have the same priority, a hash function in the BSR mechanism is used to choose the RP to ensure that all routers in the PIM-domain have the same RP for the same group.

To change the default priority of any candidate RP, use the ip pim rp-candidate IFNAME PRIORITY command. At Router_D, the show ip pim rp mapping command shows that Router_C is chosen as the RP for a specified group.

The following output displays the complete configuration at Router_C and Router_D:

This section provides the steps to verify the RP configuration.

The show ip pim rp mapping command displays the group-to-RP mapping details and displays information about RP candidates. There are two RP candidates for the group range, 224.0.0.0/4. RP Candidate 10.10.1.5 has a default priority of 192, whereas, RP Candidate 172.16.1.2 has been configured to have a priority of 2. Since RP candidate 172.16.1.2 has a higher priority, it is selected as RP for the multicast group, 224.0.0.0/4.

To display information about the RP router for a particular group, use the following command. This output displays that 172.16.1.2 has been chosen as the RP for the multicast group 224.0.1.3.

After RP information reaches all PIM routers in the domain, various state machines maintain all routing states, as a result of Join/Prune from group membership. To display information on interface details and the multicast routing table, refer to the Configuring Rendezvous Point Statically section.

To dynamically configure the RP, Router_2 on eth1 is configured as a Candidate RP using the ip pim rp-candidate command. Since no other router is configured as C-RP, Router_2 becomes the RP. Router_1 on eth1 and Router_2 on eth1 are configured as the Candidate BSRs. Since Router_1 has a higher priority value than Router_2, Router_1 becomes the BSR router and is responsible for sending group-to-RP-mapping information to all other routers in this PIM domain.

For this example, refer to Figure 3-2 for the topology.

When the ip pim unicast-bsm command is configured on an interface that is a DR for a network, then that interface unicasts the stored copy of BSM to the new or rebooting router.

The Anycast-RP feature provides load balancing among active RPs and redundancy in a PIM-SM network domain. In a PM-SM configuration, only a single active RP for each multicast group within a domain is permitted. However, in an Anycast-RP configuration, this restriction is removed with the support of multiple active RPs for each group in a domain.

OcNOS supports Anycast-RP using the PIM implementation. In PIM Anycast-RP, Multicast Source Discovery Protocol (MSDP) is not employed to share information about active sources. Instead the Register mechanism in PIM is extended to provide this same function.

The following describes Anycast-RP in PIM-SM:

Host1 and Host3 act as hosts and sources for sending join and multicast data packets; Host2 acts as a host.