Microsoft Clustering Fail-over Have Downtime
According to https://www.mssqltips.com/sqlservertip/1882/understanding-why-there-is-still-downtime-for-sql-server-with-microsoft-clustering/, Microsoft clustering fail-over solution involve the following steps that cause the downtime.
- Stops the SQL Server Agent service and any services dependent on the SQL Server service.
- Stops the SQL Server service (meaning SQL Server isn't available at this point).
- Releases NodeA's hold on the IP address and network name for VirtualNode1.
- Releases NodeA's hold on the shared storage.
- Tells NodeB to take control of the shared storage (which it does).
- Tells NodeB to take control of the IP address and network name for VirtualNode1.
- Starts the SQL Server service on NodeB (meaning SQL Server is coming back on-line for its clients).
- Starts the SQL Server Agent service and any services dependent on the SQL Server service.
I test the down time, it would be a couple of seconds. Sound little, but it would be bad enough to cause some application server (WebLogic for exmaple) 20s to 30s revcover.