6509 multicast meltdown

[MARK BAKER]

Hello all,

I have a 6509 in the core/distribution of our network. There is also a directly connected 6509 as the backup and is not currently being utilized. We have 4507R, 3750G, and 4503 switches in the closets connected to the two 6509 switches.

We are having a problem were multicast video using Microsoft media services runs fine until a large user participation of the stream is seen. Around 300 - 400 requests in around 25 subnets for the stream in a 5 minute window causing the meltdown.

At first we thought it was the load on the server, but that was ruled out today during testing. Some of the access switches receive the stream fine during this large user request, while others(most) are not receiving all the multicast packets. It seems consistant to which switches are receiving packets and which ones aren't.

All indications point to the 6509 core/distribution switch, but we still need to do further testing to determine at what level it is failing.

We have mls ip multicast running and it seems to be functioning fine through show commands. multicast seems to be working fine using show commands. The problem is that we are dropping packets to some of the access switches and it appears to be at the 6509 core/distribution switch.

If anyone has seen this before, I would appreciate your assistance.

[vmoopeung]

This can be an issue of multicast load balancing or some routing protocols.

http://www.cisco.com/en/US/tech/tk828/technologies_tech_note09186a0080094b55.shtml

[MARK BAKER]

Thanks for the response. I noticed not many multicast questions receive responses in this forum.

Turns out that MLS IP MULTICAST was causing the packetloss issue we were having when many subnets were joining the multicast group. Since, this feature is not really a benefit on our network(especially since it doesn't work), I have just disabled it and everything works fine.

NOTE: For anyone having the same issue, there were no counters in any show command that pointed this issue out. Everything seemed to be working fine on the switch. We were able to determine packetloss using packet captures.

Thanks again for your response,

Mark