NSX-T Reduced bandwidth for multicast traffic between T0 and T1

Products

• VMware NSX

Issue/Introduction

• Throughput varies over time but can be as much as 10 Gbps to 600 Mbps.
• The NSX-T TEP (Tunnel Endpoint) interfaces for the transport node (Edge node and ESXi host) are in the same rack and configured in the same subnet.
• The multicast offload ESXi host is in a different rack/subnet to the host supporting the multicast listener.
• A failover of the listener or Active T1 to another rack resolves the issue.

Environment

VMware NSX-T Data Center 3.x
VMware NSX 4.x

 

Cause

The destination MAC address for unicast traffic gets incorrectly changed from the Edge TEP MAC to the TOR MAC. This occurs when the host supporting the multicast receiver and the unicast source, receives multicast traffic from the host doing the multicast offload in another rack.

Resolution

This issue is resolved in NSX-T Data Center 3.2.3 and VMware NSX 4.1. 

Workaround:

To workaround this issue, move the Edge TEPs to a different subnet so that Host TEPs and Edge TEPs are in separate subnets.