SUPERASYNC mode

In SUPERASYNC mode, log writing and replication are independent. HADR never enters peer state.

Log shipping only uses remote catchup state. Log writing is never slowed down. However, primary-standby log gap can grow. In a failover, data in the gap is lost. You must monitor the gap closely. This mode provides the least impact on primary, at the cost of the least data protection. It is typically used on unreliable networks.

While in other sync modes, a non-forced takeover is allowed only in peer state, where primary and standby log positions are close, in the SUPERASYNC mode, non-forced takeover is allowed in remote catchup state. If there is a large gap, the takeover takes a long time because after stopping transactions on the primary, HADR is still required to ship all logs in the gap to the standby and replay them before takeover can complete. You must to check the gap before issuing a non-forced takeover in SUPERASYNC mode.