Description

This wait type is only used for databases participating in an Availability Group. When modifications occur on the Primary Replica in synchronous mode it must commit all of these on the Secondary Replicas before committing on itself. This wait type accumulates during this period

Resolved by

DBAs and Developers

Suggested solutions

1. If SLAs allow, investigate whether the Availability Group can be switched to asynchronous mode
2. Ensure that the network between replicas is not over-utilized, if so consider upgrading
3. Rewrite modification DML that performs small (row-by-row) changes. Batch modifications result in less transaction log usage and messages across the network
4. Remove unnecessary indexes, less messages and transaction log usage on modifications
5. Scale out database design

Additional research

HADRON Learning Series: HADR_SYNC_COMMIT vs WRITELOG wait
Monitor SQL Server AlwaysOn Availability Groups
MSDN Blog: Monitor Performance for AlwaysOn Availability Groups

⇐ Back to index