SD-WAN Active-Standby Cluster Disaster Recovery Latency Requirements

[klasko16]

Our team is looking to add in another management cluster in an alternate location, and registering them in an active-standby disaster recovery configuration. I am finding that intra-cluster traffic has a more strict latency requirement of <4ms while the inter-cluster latency requirement hasn't been formally established. If we have 2 NOC locations with about 25-30ms in latency between them, would we experience issues with the cluster interface/replication traffic?

[Jeongjun Park]

Hello.

I couldnt find any documents about the latency for DR.

As you saw, Cluster needs <4ms latency.

Perhaps, 25-30 ms wont be matter If link is stable.

The DR replication will be successful.

Try to open the case If you need.

Thank you.

[Joseph W. Doherty]

Cluster vendor's technical support likely best to answer that question.

On the face of it, if cluster documentation says 4 ms or less, and you have 25 to 30 ms, that's quite a delta.

Some questions that might make some difference, is the 4 ms latency one way or round trip?  Is the 4 ms for some kind of heartbeat that might be subjected to QoS to minimize queuing latency impact?  Is there any way to "tune" the value to allow for increased latency?