We have a 2 node SQL 2008 R2 cluster with 3 instances. Each clusternode has 140 Gb memory.
i've read that for performance it is best practise to set max memory settings.
I'm not a SQL guru, but i've noticed that our DBA has set a max mem on one clusternode for one instance at 50 GB, and on the second clusternode with the remanining 2 instances, max memory setting is not configured. So on the second clusternode all memory is used.
His argument was, that during a failover the memory is released, so it should not be a problem.
I disagree because of some articles i've read, but i'm not sure. What could happen during failover if node 1(instance set at 50 Gb) fails over to node 2( instances set to use all mem)?
I've read that depending on the workload of each instance you can set the max mem.
Thanx
