We are considering implementing SQL failover clustering and I have a question about how SQL Failover clustering works with regards to performance. Specifically, we would have about 1000 dbs mounted on said cluster, and my concerns are:
-
Performance during a failover: Will there be an outage during the failover operation?
-
How long can I expect the failover to take?
I'm not sure how it's implemented, but I'm just imagining the analogy of shutting down a SQL server service on one machine, and once complete starting up on another machine. With all of these dbs, just recycling the SQL service takes about 10 minutes. Should I expect the same from the failover?
Best Answer
Yes there would be some outage during cluster failover.. During failover SQL Server is stopped on current node and failedover/moved to other node. After this SQL Server is brought online on the node to which it is failed over. Sometime this is so quick that application users might not notice. But the fact that SQL Server is restarted means you should always keep in mind that there might be some outage
There are lot of factors that influence how long it would take for failover cluster to come online on other node. The factors are (but not limited to)
Edition
of SQL Server. Fast recovery which isenterprise edition
feature could bring database online after second phase(redo) of recovery