We have a very outdated SQL cluster on Windows 2k using VCS for Win 2.0 on HP DL580 G1's. Recently there have been some severe performance issues with SQL. We are on a path to upgrade once the dollars come through. The servers sit around 90% CPU and 98% memory usage. There are 4 SQL DB instances on each cluster server. In the event of a cluster failover, how would the resources on the receiving server re-act. I see it several different ways.
A: The failover occurs, and resources balance out (optimal).
B: The failover won't occur because there aren't enough resources.
C: The failover occurs, but because of resource issues, now both servers in the cluster crash.
I know we desperately need to upgrade the entire stack, but that is management decision and out of my control. I'm just wondering that since we are configured with clustering, would we be better off disabling it as to not take down both servers in the cluster in the event of 1 server failure.
MSSQL likes to suck up as much free memory as it can, so the memory usage wouldn't bother me at all. The CPU usage is high, but having the resources from two nodes running on a single node would only slow SQL down I think, rather than actually cause any issues.
If you have a quiet period, the best way to find out might be to simply switch the resources all on to the same node.