Can I join a 3rd node to a windows 2012 cluster if the first 2 are using a shared drive?

I am in a situation where we started DR as an after thought. We built our 2 node Windows 2012 cluster, then installed SQL 2012 with alwaysOn and that is working great. However now the company wants to build a data center on the other coast. We have all the network magic in place, I can copy files from node1 (lives in CA) to node 3 (lives in PA), I can connect from SSMS to the instance on node1 all that good stuff. However I am wondering if I can just join node 3 to the existing windows fail over cluster. We don't have shared disk between node1 and node3 and the original windows cluster was built using shared disk between node1 and 2. I was thinking aobut just trying it, but this is a production server, and I don't want to take any unnecessary risks if I can help it.

So....Has anybody out there been in this situation? Any advice?



For a stretched cluster across sites (which has its own significant challenges), you'll want to use a UNC share as a witness, not a disk. But yes, such a change can be done.



For a stretched cluster across sites (which has its own significant challenges), you'll want to use a UNC share as a witness, not a disk. But yes, such a change can be done.



Unless your subnets span datacenters, you'll also have to move to a multisubnet deployment. So you'll want to read up on that a bit as there are some intricacies around behavior and connection string changes for the connecting to the listeners. Luckily once you have it added to the WSFC, you can test all that stuff out by just building a new AG with the 3 replicas and test all of that before migrating the databases into that AG. I have a few resources for multisubnet deployments, which I'd be happy to share if you are indeed moving to that configuration.



Thanks experts! I will be cr eating the AAG to be "stretched" across the 3 nodes. In the event the CA Datacenter goes completely offline we want the ability to run our database from PA without having to modify the application connection strings. My fear was that since the first 2 nodes in the windows cluster were using shared disk that adding a 3rd node using a witness would not be possible. I will take the resources please :)



Listeners - http://blogs.msdn.com/b/sqlalwayson/archive/2012/01/16/introduction-to-the-availability-group-listener.aspx - I include this because I have never had a listener configuration in a multisubnet deployment go smoothly

AlwaysOn in general (this is largely the top of the AlwaysOn BoL - which does have terrific documentation on this, but you often times have to look fairly hard for multisubnet variations) - http://msdn.microsoft.com/en-us/library/ff878487.aspx



Thanks!

Share this

Related Posts

There was an error in this gadget