Core Issue
Stub networks are those segments that receivers are directly attached to for any multicast group, even though there are no receivers interested in multicast traffic beyond that segment. Stub multicast routing allows such stub networks, or sites, to be configured easily for multicast connectivity. It is useful when the stub network is connected to the central site with slow speed links.
This is useful in Protocol Independent Multicast (PIM) dense mode where periodic flooding and subsequent pruning of multicast traffic occurs for unwanted multicast groups. Stub multicast routing prevents periodic flooding and pruning, while at the same time allowing multicast traffic to be forwarded for groups in which receivers are available on the stub network. This can work even when there are multicast sources available on the stub network sending traffic to receivers connected through the central site.
Stub multicast routing is also useful in PIM sparse mode environments. This eliminates the need to maintain the group-to-RP mapping cache on the stub router. This also saves periodic update bandwidth if Auto-RP or PIM Bootstrap Router (BSR) is used for distributing the Rendezvous Point (RP) information. However, this is possible only when receivers are available. It is not possible when multicast sources are attached to the stub network because the stub router functioning as the first-hop router needs to register with the RP.
Resolution
To configure stub multicast routing, issue the ip multicast-routing command globally, and configure ip pim sparse-dense-mode commands respectively on the interfaces on both the central site as well as the stub routers. This is necessary for routers to forward the received multicast traffic from one interface to the other.
On the stub router, issue the ip igmp helper-address x.x.x.x command on the interface on which the receivers are connected, where the x.x.x.x address refers to the address of the upstream central site router that is one hop away. This command enables the stub routers to act as Internet Group Management Protocol (IGMP) proxy agents instead of fully participating in PIM, and relays the IGMP messages from the receivers to the upstream router. The central site router therefore treats the stub router as a directly connected receiver for a group, as long as there is a receiver for that group available on the stub network.
On the central site router, issue the ip pim neighbor-filter y command on the interface on which the stub router is connected, where y refers to a standard Access Control List (ACL) that has a deny statement for the address of the stub router. This command filters all the PIM control messages from the stub router and prevents treating it as a PIM neighbor. This filter is also used to control and form neighbor relationships only with trusted routers.
In dense mode, when multicast traffic arrives at the central site router for a group for which there is a member available on the stub network, the central site router forwards it to the stub router. Multicast traffic is forwarded because it is treated as an attached IGMP host for the proxy messages relayed to it. However, when multicast traffic arrives at the central site router for a group for which there is no member available on the stub network, the central site router prunes the traffic. This is because the stub router is not treated as a PIM neighbor due to the configured filter. The interface connected to the stub router is never added to the outgoing interface list for those groups.
In sparse mode, when the stub router relays the IGMP messages to the central site router, a PIM Join message is sent towards the RP by the central site router to join the shared tree and pulls the traffic. The traffic flows over the shared tree to the stub network, even with the stub router not knowing about the RP-to-group mapping.
For more information on stub multicast routing, refer to these documents:
