May 6, 2009 at 6:46 am
What are your thoughts on adding another sql server instance to a cluster node, for example in a 3 node cluster with single instances on the 3 nodes or on a active-passive cluster with default instance.
May 7, 2009 at 6:05 am
Anyone?
May 8, 2009 at 4:14 am
May be this documentation usefull for you:
May 8, 2009 at 7:14 am
That link is for adding or removing nodes in SQL 2008 Cluster.
--
All cluster nodes that I have worked on had single sql server instances. Has anyone added another instance in a clustered environment & what are their thoughts on that???
May 8, 2009 at 8:15 am
Google for Active-Active SQL Server Cluster
* Noel
May 8, 2009 at 8:26 am
I've worked with active - passive, active - active, active - active - active and active - active - active - passive, sorry to labour the point a bit but these were with different companies and they all used clustering in different ways and to be honest the only thing that I think was the deciding factor when people set these up were how many resources were available and how the person who set the clusters up interpreted or understood how they worked or the options available.
To me they all work as well as each other, the servers are in a cluster so if one goes down then the services fail over, I think the best configurations are where all the nodes are active so you use all the resources the best you can but this also causes problems if you want to take one node out of the cluster for maintenance for whatever reason and that would mean you had no failover for the other node in the cluster. In a similar way if you have several active nodes and one passive for the failover, if that passive node goes down you have no failover for all the other nodes so it really is down to what risks you are willing to accept and how many servers you have available, if you have the resource then active - passive is the way to go.
Hope this makes some sense and it helps.
May 8, 2009 at 8:58 am
It's all about the resources on the box(s) - Disk, Memory, cpu.
AND if there are enough resources on a single node to handle a failover if that were to occur.
I've run multiple instances on a cluster node in a couple different shops with no problems.
Tim White
Viewing 7 posts - 1 through 6 (of 6 total)
You must be logged in to reply to this topic. Login to reply