A SQL virtual server is the equivalent of a cluster resource group. If a resource within a virtual server fails, the entire group and all of its resources failover to the second cluster node. Each SQL virtual server must contain the following:
- A Disk resource (each virtual server requires its own disk volume)
- A Network Name resource
- An IP Address resource
- One instance of SQL 2000 per virtual server
Note: Once an instance of SQL Server is installed, resources are added to the virtual server. These resources include a SQL Server resource, a SQL Server Agent resource, and a Fulltext resource.
Preliminary steps need to be completed before beginning the installation of SQL Server in a cluster environment. Windows 2000 Advanced Server (or other version of Windows) and Cluster Service need to be installed and configured. You need to verify that there is a Microsoft Data Transaction Coordinator (MS-DTC) resource created through Cluster Administrator.
SQL requires the MS-DTC for distributed queries. It is installed be default when Windows 2000 is installed, but when in a cluster configuration, the MS-DTC needs to be run in cluster mode. This component can be put in cluster mode using the Component Cluster Wizard (Comclust.exe). When the executable is run, it makes the following changes:
- Creates an MS-DTC resource
- Creates an MS-DTC log file on one of the shared disks (this ensures the log file can be accessed from any node)
- Places the necessary MS-DTC registry entries in the cluster registry
Once configured in cluster mode, the MS-DTC runs on a single node, which means that only one transaction manager is in control of the cluster at any given time. Any calls made by the other cluster nodes are forwarded to the transaction manager that is in control of the cluster. If the MS-DTC fails, it is restarted on another node in the cluster.
To configure the MS-DTC to run in cluster mode:
- Click Start and click Run.
- Type cmd and click OK.
- From the command prompt window, type comclust.exe and press Enter.
- Repeat these steps on each node in the cluster.