Hi Deepak Kumar
From what I can see, the main difference between RBKC and SW-VM02 is that RBKC is happily running over UDP multipath, while SW-VM02 is falling back to WebSocket, which explains the higher round-trip time and degraded performance. The packet captures showing resets from 10.0.0.4 to 172.26.36.50 suggest that UDP traffic isn’t being allowed through somewhere along the path, so the session defaults to TCP/WebSocket. That’s why you’re seeing 320 ms latency and low bandwidth on SW-VM02 compared to the 200 ms and higher throughput on RBKC.
The fix usually comes down to making sure UDP 3478–3481 is open end-to-end (client, firewall, NSG, and any intermediate appliance). Double-check that your NSGs and firewalls in UK West are aligned with the RBKC setup in UK South. Also, confirm that the AVD session host in SW-VM02 has the right outbound rules and isn’t blocking UDP itself. Once UDP is properly enabled, the client should negotiate multipath again, and you’ll see the round-trip time drop significantly. If you still see resets after that, I’d recommend running a targeted trace on those ports to confirm whether the block is happening at the network edge or inside the VM.
So in short: align the UDP rules with RBKC, validate NSG/firewall configs, and re-test the session. That should get SW-VM02 back on track.