Auto-tuning memory configuration on a cluster 
(Merrill Aldrich) I was inspired by Aaron's post on a similar topic a while back to dig deeper into how I could automagically adjust the memory allocation for multiple instances of SQL Server in a cluster-failover scenario. So, suppose you have:
--A two-node cluster
--Four SQL Server instances, two intended to run on each physical server (call it "deluxe active-active")
--An environment where it's feasible to run all four instances on one node, in case of an emergency or in a maintenance window
--An expectation that some performance degredation is acceptable in the emergency scenario, and you want to take full advantage of your hardware investment when both nodes are up and running
--Stuff running that is tolerant of reconfiguration of max memory in SQL Server "on the fly"
Read Article >>