I was going to create a Windows Server Failover Cluster, in order to set up MS SQL Server AlwaysOn Availability Groups.
As a prerequisite of failover clustering, I've created and configured an Active Directory using two servers. One server is domain controller, another server is the member of the domain.
While following all the steps and suggestions of creating a failover clustering, I've found that all the nodes should be in the same Domain Role. As long as I understand, the domain must have at least one domain controller. And in order to create a meaningful Failover Cluster, there should be 2 or more nodes. This means that I need three server to setup a basic failover clustering with 2 nodes? One is domain controller, which is not going to participate in the clustering, and 2 others are domain members and are nodes of the failover cluster?
On other article, about Sql Server AlwaysOn availability group perquisites, the system/server/computer should not be a domain controller, but all the nodes should be domain members.
Ok, then I need three servers to setup two node (one primary, and one secondary replica) Sql Server AlwaysOn Availability Groups on top of a two node Failover Clusters (two domain members, and 1 domain controller)? How is that? Or I'm missing something, or it can be configured with only tow servers somehow?
Thanks.
Best Answer
There is some ambiguity in your question, but the following presumes you are talking about setting up a high-availability MS SQL environment on a Hyper-V failover cluster.
You are correct in so far as you will need a domain controller separate to the failover cluster nodes - for example, to reliably authenticate to the cluster nodes (some sysadmins set up a system in which the only domain controllers are VMs hosted on the failover cluster but this is inadvisable for fairly obvious reasons).
So your basic setup will be:
Domain controller on a separate server (can be run as a VM or on a physical server)
Failover cluster node 1
Failover cluster node 2
Now, your failover cluster nodes will share the role of hosting a bunch of VMs of your choosing, which you can set up after running through the failover clustering wizard in adding server roles to the physical servers. Two such VMs should be running MS SQL instances, which you can then configure for high availability as per the high availability guidelines you have. It is a good idea to distribute VMs which share roles between your cluster nodes, to minimise or completely mitigate any downtime if one of the nodes were to unexpectedly go offline at any point.