Long merge and Slow merge SQL Server Replication Warnings
Originally published on KendalVanDyke.com.
During my presentation about Transactional Replication at SQLSaturday #8 – Orlando 2008, I was asked what the difference was between Long merge and Slow merge SQL Server replication warnings. I didn’t know the answer (it was a session about transactional replication after all), but not being one to let a good question go unanswered I had to find out.
Here’s what they mean:
- Long merge refers to the amount of time that the merge agent has run (or is currently running) during the most recent synchronization. If this value exceeds the set threshold the alert will fire.
- Slow merge refers to the number of rows processed per second by the merge agent while running. If this value falls below the set threshold the alert will fire.
There are two threshold values for each alert – one for dialup connections and one for LAN connections. For dialup the default is 8 minutes for long merge and 20 rows/sec for slow merge. For LAN the default is 3 minutes for long merge and 100 rows/sec for slow merge.
Thresholds are enabled and set in Replication Monitor by selecting your merge publication in the left pane and the “Warnings” tab in the right pane. Alert actions are set by clicking the “Configure Alerts ” button in the same pane or in Management Studio’s Object Explorer under SQL Server Agent –> Alerts. A screenshot of the Warnings tab in Replication Monitor is below.