High commit latency probe

Using FoundationDB
1

Hi all! I’m trying configure FDB for deployment in cloud.
I have:

  • 3 availability zone, with latency between each 20ms.
  • 2 machine C4-16G = 1 x logs + 1 x storage + 2 x stateless in each zone
  • triple redundancy mode

As result status return commit latency probe equal 20ms. If I deploy only in one zone, it will be 1-2ms.

How should I configure deployment for all zones and reduce latency?

2

Mode ‘triple’ means log replicas are three, only when three log replicas are all written to each zone then the commit is successful. so it would be 20ms at least.

3

Do I understand correctly that only a multi-region configuration can reduce latency? In this case, the solution is to use a single availability zone.

From manual:

FoundationDB attempts to replicate data across three datacenters and will stay up with only two available. Data is replicated 6 times. Transaction logs are stored in the same configuration as the three_data_hall mode, so commit latencies are tied to the latency between datacenters.

Are all redundancy modes dependent on latency between zones/halls/DCs, and is async replication only available for a multi-region configuration?

4

Deploy your servers in a single availability zone and choose double/triple redundancy unless you need a higher redundancy level.

Are all redundancy modes dependent on latency between zones/halls/DCs, and is async replication only available for a multi-region configuration?

A higher level of redundancy will distribute replicas across DCs/regions, so the latency will be higher.

Async replication can avoid high latency between regions and failover, and it looks like satellite/satellite_logs/satellite_logs can only be configured in multi-region redundancy.

1 Like