To increase its availability, SQL 2000 supports failover clustering, which means that processing can automatically switch between cluster nodes if one fails. For a corporation, this means that in the event of failure, database services can quickly be restored. As with Exchange, certain components of SQL Server are supported in failover clustering. The supported components include:
- Microsoft Search Service
- Multiple Instances
- SQL Server Enterprise Manager
- Service Control Manager
- SQL Profiler
- SQL Analyzer
When configuring failover clustering for SQL Server, you have two options. The first option is to install SQL Server on all cluster nodes. For each installation of SQL, a virtual server is created. The second option is to configure multiple instances of SQL Server (with SQL 2000 you can install up to a maximum of 16 instances, each operating as though it were on a separate server). Each instance that is installed has its own path for executable program files and its own location for storing data files. Every instance will have its own database and database files that are not shared between instances. When an instance of SQL is installed, the setup program automatically installs the needed executable files on both the cluster nodes (the executable files are installed in the same location on a cluster nodes) and places the database files in a cluster group (virtual server). The cluster node where the instance is installed then becomes the owner of the cluster group. In the event of failure, only the database will failover, not the executable files, because they are already located on each cluster node.
Two types of instances can be configured on cluster nodes, as explained in the following list:
- Default instance – When SQL is installed for the first time on a cluster node, it is automatically installed as a default instance (this can be changed by deselecting the appropriate option during setup). You can enable client application to connect to the default instance by specifying the network name of the computer (this obviously means that there can only be one default instance running on a computer).
- Named instance – Multiple instances of SQL Server can exist on a single node when you create named instances. You can enable client applications to connect to a named instance by specifying the computer name as well as the instance name (this is the main difference between a default instance and a named instance).
Note: Keep in mind that each instance of SQL Server must have a unique name between cluster nodes to avoid conflict.