In the previous posts in this series, I discussed how a SAN, Gigabit switching, and host servers are the three pieces needed to do Hyper-V clustering. This article will continue that discussion by covering host setup, and cluster shared volumes.
Windows Server 2008 version
The first item to consider is which Windows Server 2008 licensing to use. If you take a look at Microsoft ’s comparison of editions for R2, you’ll see that Standard, Enterprise, and Datacenter editions each come with a different number of virtual machine instances included. Standard edition does not support Failover Clustering, so make sure you have Enterprise or Datacenter. Enterprise includes rights to run up to 4 virtual machines. Datacenter includes rights to run an unlimited number of virtual machines.
|Edition||Virtual Machine rights||Failover Clustering|
Basic Host Setup
Preparing the hosts to support clustered virtual machines involves a few basic steps.
- Add hosts to your domain – Other Hyper-V configurations do not require that the host be a domain member. However, to support clustering, the hosts must be domain members.
- Install Hyper-V Role – At this point, the hosts could run virtual machines. However, they cannot yet be made highly available.
- Install Failover Clustering Feature – This enables clustering on your server. One of the cluster objects that can be created is a clustered virtual machine.
Server Manager with Hyper-V and Failover Clustering installed
Now you should make sure that all network connections are in place are ready to go.
- Client Connectivity – this connection should be on a subnet that is accessible by client computers and will also have access to Active Directory. For discussion, let’s say the host will connect to Active Directory using subnet 10.10.10.0 with subnet mask 255.255.255.0.
- Heartbeat – In a two-node cluster, this could be a direct connection. If you have more nodes, or want to build with scalability in mind, a switch should be used. For this one, let’s use 192.168.1.0 with subnet mask 255.255.255.0.
- SAN Connection – This connection needs to provide Gigabit speed or faster between the hosts and the SAN. For this one, let’s use 172.16.1.0 with subnet mask 255.255.255.0.
Cluster Shared Volumes
Before R2, Windows Server 2008 required a separate disk instance for each virtual disk that virtual machines used. If you had a total of ten servers using 16 virtual disks, then you would end up with 16 disks being used. Since the operating system drive for each of those had to use a drive letter, pretty soon you had alphabet soup in Windows Explorer and Disk Management. Cluster Shared Volumes (CSV) has been added to streamline administration in R2 (aka Hyper-V 2.0). A CSV can now support the files and access for more than one VM. In a small environment, this could mean that all VMs use one CSV total or one CSV per host.
When you enable Cluster Shared Volumes in the Failover Clustering console, the C:\ClusterStorage directory is created which will contain a folder for each CSV in the cluster. This takes care of the alphabet soup I mentioned earlier. Using cluster shared volumes is not a requirement but I would highly recommend its use. There are some special requirements for using cluster shared volumes that are well documented at Requirements for Using Cluster Shared Volumes.
In this article I covered which versions of Windows Server you should use for clustering. I also covered getting the hosts ready and cluster shared volumes. In the next article in this series, I’ll be covering quorum and disk configuration.