February 21, 2012 at 1:54 pm
Is it possible to move current SQL 2008 Standard Edition 64 bit Cluster to new hardware wthout down time? The customer has purchased new hardware and would like to migrate their current cluster to the new hardware. Below are the steps we would like to follow.
1. Install exact same OS on current cluster to newserver_2 and newserver_1
2. Remove oldserver_2 from existing cluster
3. Shut down oldserver_2
4. Add newserver_2 to cluster
5. Failover sql to newserver_2
6. Shutdown oldserver_1
6. Add newserver_1 to cluster
7. Failover sql to newserver_1
Is this possible?
February 21, 2012 at 2:26 pm
At a high level I believe it will work. I don't believe the nodes have to be evicted from the cluster for it to work just that SQL Standard cannot have more than 2 possible nodes. So removing the node from SQL is the important bit.
I remember testing this last year with a three node cluster. SQL Standard installed fine in the clustered environment and didn't balk until I tried to add the third node.
CEWII
February 21, 2012 at 2:38 pm
Elliott Whitlow (2/21/2012)
At a high level I believe it will work. I don't believe the nodes have to be evicted from the cluster for it to work just that SQL Standard cannot have more than 2 possible nodes. So removing the node from SQL is the important bit.I remember testing this last year with a three node cluster. SQL Standard installed fine in the clustered environment and didn't balk until I tried to add the third node.
CEWII
Are you saying that we could?
1. Add the new nodes to the Windows cluster.
2. Remove one of the old nodes from the evict one og the old nodes
3. Add one new node to the sql cluster
4. Evict the last old node
5. Add the last new node to the sql cluster
February 21, 2012 at 2:50 pm
I'm saying that you could have the new nodes and the old nodes as part of the cluster at the same time. What matters is the number of machines that SQL is configured to run on. In the case of standard it cannot be more than two. So you would use the remove option from the SQL setup and take the passive node out of the SQL cluster (the node would still be in the overall cluster). Remember you have an overall cluster which defines a set of related machines and a clustered SQL instance that resides on the cluster. You are taking one of the nodes out of the SQL Instance. At that point it will only show one possible node (not preferred but possible node). At this point you can add one of the new nodes to the SQL cluster. When that is done you can failover to the new node. And when you are happy with the new node you can remove the last old node from the SQL instance. At this point you are on the new hardware for SQL. Add then second new node to SQL. At this point you are able to failover between the new hardware and not at all to the old because the old hardware is not possible nodes anymore. Once you are sure of the new hardware you can evict the two old nodes. At no point in the process can the possible node count rise above two, and this process follows that.
CEWII
February 21, 2012 at 2:53 pm
Sounds good. Thank you sir!!
Viewing 5 posts - 1 through 4 (of 4 total)
You must be logged in to reply to this topic. Login to reply