When a high availability syslog-ng Store Box (SSB) cluster is operating in a high-latency environment or during brief periods of high load, there is a risk of slowness, latency or package loss. To manage this, you can compensate latency with asynchronous data replication.

Asynchronous data replication is a method where local write operations on the primary node are considered complete when the local disk write is finished and the replication packet is placed in the local TCP send buffer. It does not impact application performance, and tolerates network latency, allowing the use of physically distant storage nodes. However, because data is replicated at some point after local acknowledgement, the remote storage nodes are slightly out of step: if the local node at the primary data center breaks down, data loss occurs.

To turn asynchronous data replication on, navigate to Basic Settings > High Availability, and enable DRBD asynchronous mode. You have to reboot the cluster (click Reboot cluster) for the change to take effect.

Under prolonged heavy load, asynchronous data replication might not be able to compensate for latency or for high packet loss ratio (over 1%). In this situation, stopping the slave machine is recommended to avoid data loss at the temporary expense of redundancy.