The HA design we propose in a nutshell:
Primary site A
Node 1: 1 x physical host for a SQL Failover cluster instance for application A
Node 2: 1 x physical host for a SQL Failover cluster instance for application B
Node 3: 1 x physical host for two SQL Failover cluster instances, one for a test env for application A, and another for a test env for application B
Secondary site B
Node 4: 1 x physical host for a SQL Failover cluster instance that acts as a 3+1 failover node that will pic up the failure of either or all of the 3 nodes in Primary site A.
Now I am a complete novice, but my understanding of quorums are that their purpose is to establish when the cluster must stop running due to things like network failures etc. There really is not much out there in terms of a laymans description for quorums
The basic requirement for a multi-site failover cluster with an even number of nodes is to use the node & file share majority configuration, and that is what I have here. 4 x nodes (3 at site A and 1 at site B), and a seperate file share for the witness
file (which makes the odd number for a majority).
So my understanding is that:
If node 1 fails, it will failover to node 4
If node 1 and 2 fail, they will both failover to node 4
If node 1, 2 and 3 fail, they will all fail
View Complete Post