Multi PoP deployments

When PoPs not in same L2 domain, there can be issues for control traffic (i.e,That is IPv6 communication between E2E controller and nodes).

  1. When a POP is down, BGP router will deliver Downlink control traffic of that prefix zone’s nodes to any of the other POPs in random (lets call it POPx). If there is no wireless patch between POPx and certain nodes, those will show up as offline.

  2. Each POP advertises its own zone prefix and seed prefix to the BGP router. When multiple BGP routers are in play, please make sure IPv6 traffic for a particular zone gets routed to right POP.

We have a mult-pop deployment with 3 pop nodes, with an on-premises E2E controller, we are also using L2 tunnelling. Two of the pop nodes have BGP peering sessions with one router and the other pop node peers with a different router. We have noticed that when one of the pop nodes reboots, or drops its wireless link to the rest of the network, this causes a lot of instability in the rest of the network for several minutes, with end nodes changing their layer-2 tunnel endpoints a couple of times, we also sometimes see wireless link in the middle of the network drop and re-establish when this happens. I was just wondering if these issues might be related to BGP session timeouts and related routing instability with the connection to the E2E controller? If this is the case, then would you have any recommendations for setting the BGP keepalive and hold timers for the BGP sessions on our router that might help with this?

Thank you for reaching out to us with your concerns.
Absolutely it is wise to consider multiple factors contributing to network instability.

Based on the information provided one potential factor could be related to BGP.
Recommend to set the HOLD time in router to be 240s and at the POP we can increase the BGP keep-alive timer to 80s.
Feel free to to experiment with different hold timer values to find optimal setting for your network.

It is expected for tunnels to reroute to a reachable POP and then revert to their zone POP once it reboots.This ensures network redundancy and high availability.
However to note these reroutings should not affect the wireless links.

If instability persists in your network please raise a support ticket and we will be more than happy to investigate and work towards increasing the robustness of your network.