IP Multicast in a Campus Network
IP Multicast Large Campus Design
Each client and server access-layer switch is dual-connected to, and served by, a pair of distribution-layer routers running HSRP. For multicast, one of the two routers is the DR with the other being the IGMP querier. The IP Unicast routing protocol is configured such that the trunk from the access-layer switch to the DR is always preferred over that of the non-DR. This forces the unicast and multicast paths to be the same. The IGMP querier is responsible for sending IGMP queries. Both the DR and the non-DR receive the subsequent reports from clients, but only the DR will act on them. If the DR fails, the non-DR will take its role. If the IGMP querier fails, the DR will take over its role. The distribution routers have dual connections to the core.
Keep the RP placement simple. With Anycast RP, it is recommended that the RPs be placed in the center of the network. Placing RP operations on the core-layer switches is a good idea because the core is the central point in the network servicing the distribution-layer switches in each building, the aggregation-layer switches in the server farms, and the WAN and Internet blocks.
The applications used in this sample design (MoH and IP/TV) have few sources to many receivers. The sources are located in the server farm. So, a complex distribution of RPs throughout the network is not required.
Due to the number of devices in a large campus design, this section presents configurations for a sampling of the devices. The remaining devices should have similar configurations with the exception of the unique IP addresses, VLANs, HSRP, and other specifics.
In this design: The access layer switches have IGMP snooping enabled. The RPs are located on the two core-layer switches. PIM-SM is configured on all distribution-layer switches and core-layer switches. Anycast RP is configured for fast recovery of IP multicast traffic. PIM-SM and MSDP are enabled on all core-layer switches. Each distribution-layer switch points to the Anycast RP address as their RP. MSDP is used to synchronize SA state between the core switches.
Only two RPs are required to run Anycast RP. In most situations, two RPs will sufficiently provide redundancy for multicast. The following sections show the RP configuration for “6k-botleft-core” and “6k-botright-core”.
Core-Layer Switch Configuration
The following example shows the multicast configuration for “6k-botleft-core.”
ip multicast-routing ! interface Loopback0 description MSDP local peer address ip address 10.6.1.1 255.255.255.255 ! interface Loopback1 description Anycast RP address ip address 10.6.2.1 255.255.255.255
Cisco AVVID Network Infrastructure IP Multicast Design