I've never configured a cluster, so I'm trying to fully understand something that's bugging me - how do you eliminate your SAN as your single point of failure in a cluster?
This caught my interest: http://windowsitpro.com/windows-server-2012/windows-server-2012-building-two-node-failover-cluster
I'm trying to conceptually understand how this scenario would work...
For maximum reliability, I want to cluster an application that runs on SQL server. If I understand it correctly, I need at least two servers and one SAN. For increased reliability, I should have two switches, and each node running SQL would have two connections - one to each switch. So far, this is pretty redundant...but what about the SAN itself.
If you want to run any given application in a cluster and have it fail over to another node, do you (in a "perfect" world) need two SANs as well? If so, how do you cluster those? Does the application automatically fail over to the live SAN?
Thank you,
Ed