A network engineer observes a spike in controller CPU overhead and overall network utilization after multicast is enabled on a controller with 500 APs. Which feature corrects the issue?
The controller supports two ways of delivering multicast to APs. In unicast mode, the controller replicates each multicast packet and sends a separate CAPWAP-encapsulated copy to every AP. With 500 APs that means 500 copies of every packet, which explains the observed increase in controller CPU load and in network utilisation. In multicast mode, the controller sends one packet to a CAPWAP multicast group address. The network infrastructure replicates the packet only where required, and all APs that joined the group receive it, so the replication work moves from the controller to the network. The remedy for the symptoms is therefore to configure the multicast mode, with a multicast group address, for the AP multicast delivery method. IGMP snooping and broadcast forwarding control which clients or VLANs receive the traffic, but they do not change the per-AP replication cost, and the unicast mode is the cause of the issue. After the change, verify the configured AP multicast mode and the CAPWAP multicast group address in the controller's multicast settings, and confirm that the network routes that group to the APs.
[Reference:Blueprint 3.1 and 3.3 (multicast components and multicast on a WLAN); Cisco Wireless Controller Configuration Guide, multicast modes. Source key listed D; corrected to B (see appendix)., ]