In my organization we stood up a test SQL failover cluster. We began with the typical 2 node architecture where node1 holds the SQL role and node2 is configured as an additional node on standby in the SQL cluster. When we write big blocks of data into the test data base and pull the plug on node1 the cluster gracefully hands the SQL role to node2 and the data being written continues to write. Great - that's what we want!
Next - we decided to get fancy. Instead of installing the SQL role on one server and then adding the other node to the SQL server failover cluster, we installed a new SQL Server failover on both nodes. Next we ran the SQL installer and added node1 to node2's sql instance and vise versa for node2. Now we have two SQL server roles - with a preferred owner for both:
- SQLROLE1 -- preferred owner is node1
- SQLROLE2 -- preferred owner is node2
We can successfully run both roles off of one node - however when failing over the database write is interrupted (which basically defeats the purpose of the failover cluster)...Anyone have some thoughts on how we can get the same write performance we were having before with an A B config versus the A A (Multi-Instance) config?
Thanks,
Justin
Justin Carlton