Background: Resilient Ethernet Protocol (REP) on Cisco IE5000 switches with IGMP snooping enabled on all switches in segment. When a fault is introduced into the REP segment, multicast traffic is interrupted for a period of approximately 4 seconds. I have narrowed it down to the IGMP snooping that is causing the disruption, learning/rebuilding etc. This is layer 2 multicast traffic. Two queriers exist on the REP segment as it stands, however, I have tried a few combinations of queriers.
Unicast traffic is only disrupted for a period of approximately 50ms.
If IGMP is disabled ring wide, no multicast interruption experienced.
Efforts: Attempted to play with IGMP timers to no avail.
Questions: Is it good practice to have IGMP enabled on a REP segment? Is there a way to improve multicasts convergence speed over a REP segment.
No comments:
Post a Comment