Question
I have recently set up a two-node cluster using Windows 2000. I have a single RAID 5 shared disk array that is sliced into two logical disks which appear in Windows 2000 Disk Manager as Disk 1 and Disk 2. I have also created one primary partition on each disk and have assigned drive letters of Q: & W: respectively. I have used Drive Q: for the Quorum Resource and W: for the data files.
I have successfully installed one default instance of SQL Server on the cluster. My intention is to run multiple instances of SQL Server on the cluster in an Active/Active configuration. However, when I try to installing the second instance of SQL Server, I find that Disk W: is no longer available as a disk resource, and that the only disk available is Q: How can I create more than one physical disk resource using the same disk? Or, do I have to slice the array further and use logical disks 3,4,5 and so on?
Answer
In an Active/Active configuration, each instance of SQL Server must have its own dedicated shared array space, that is why you are seeing the problem you are.
If you want to run more than one instance of SQL Server in an Active/Active configuration, you must have at least two shared disk arrays available, one for each instance of SQL Server. These disk arrays must be physical, not logical. This is because they must run independently of each in order for failover to occur properly. You can’t create more than one physical drive out of one physical drive.
You have no choice but to reconfigure your current array so that you end up with at least two physical arrays, or you must get more physical drives.
In addition, the Quorum drive should also be on its own physical disk array, which means you should have at least three shared physical disk arrays for your Active/Active SQL Server clustering configuration.
]]>