How can Cisco 4500X send IGMP join message to the multicast source without high CPU usage?

[Laszlo Zoltan]

Our multicast service provider need IGMP join packet in order to send the multicast traffic. The service provider does not support PIM routing. We have a Cisco 4500X switch between the multicast service provider and the multicast receiver.

If I set on the upstream interface of the switch the “ip igmp join-group xxxx” then the CPU usage goes high because the traffic is also punted to the CPU. (There are lot of multicast streams.)

Are there any method to send IGMP join message to the upstream interface of Cisco 4500X switch without high CPU utilization?

Thanks,

Laszlo

[Jaderson Pessoa]

Hello,

Here there is a good topic about the same issue: https://community.cisco.com/t5/switching/multicast-causing-high-cpu/td-p/1043939

[Georg Pauwen]

Hello,

on a side note, the 'ip igmp join-group' will indeed process switch the multicast packets, as this causes the router itself to become a receiver for the multicast streams. An alternative would be to use 'ip igmp static-group', which will cause traffic to be fast switched.

Check the document below for details:

https://www.cisco.com/c/en/us/support/docs/ip/ip-multicast/119383-technote-ip-multicast-00.html

[Laszlo Zoltan]

Thank you. I have read this document.

As my knowledge the switch does not send "IGMP join" message for the command "ip igmp static-group".

We do not need that the router itself becomes a receiver for the multicast streams. We need to send "IGMP join" message to the multicast stream provider. They send the multicast stream after receiving "IGMP join" message only. They do not support PIM

How can we get 4500X to send "IGMP join" message to the multicast stream provider without high CPU usage on the switch?

Thanks,

Laszlo

[Laszlo Zoltan · Accepted solution]

I could solve this problem using "ip igmp proxy-service".

On the downlink interface I set these:

!

interface Vlan2

 description Downlink

 ip pim passive
 ip igmp static-group <StreamGroupAddress> source <StreamSourceIP>
 ip igmp mroute-proxy Vlan3

!

On the uplink (to the multicast stream provider) interface:

!

interface Vlan3

 ip pim sparse-dense-mode
 ip igmp helper-address <ServiceProviderRouterIP>
 ip igmp query-interval 15
 ip igmp proxy-service

!

How does it work?

On Interface Vlan2:

-The "igmp static-group" puts the multicast group address into the mroute table.

-The "ip igmp mroute-proxy Vlan3" sends mroute data to the Vlan3.

On the Vlan3:

-The "ip igmp proxy-service" enable the interface to serve the "igmp mroute-proxy" request.

-The "ip igmp helper-address" sends "igmp join" to the uplink router (multicast stream provider).

-The "ip igmp query-interval" sends igmp message periodically.

Using these settings the 4500X sends regularly the IGMP join message to the multicast provider while the CPU load on 4500X is normal.

Regards,

Laszlo