Failover Clustering Example
The following example illustrates how to configure Microsoft SQL Server 2005 failover clustering.
CLUSTERNODEA and CLUSTERNODEB are two computers in a failover cluster.
Run SQL Server Setup on the cluster node in control of the cluster disk where SQL Server data files will be located, and create a default instance of SQL Server on a failover cluster instance named SQLCLUSTA. This failover cluster instance can run on either CLUSTERNODEA or CLUSTERNODEB.
From this point forward, connect to the server by specifying SQLCLUSTA as the server name in the connection string.
Run Setup again on the cluster node in control of the cluster disk where SQL Server data files will be located.
|The cluster disk used by SQL Server must appear as a separate disk. You must not use the logical partition of a disk used for a previous failover cluster installation.|
Create a named instance, Inst1, on a new MSCS server named SQLCLUSTB. This must be installed in a different Microsoft Cluster Service (MSCS) cluster resource group. This failover cluster instance can run on either CLUSTERNODEA or CLUSTERNODEB, as defined by the nodes you added to the cluster definition during SQL Server Setup.
From this point forward, connect to the server by specifying SQLCLUSTB\Inst1 as the connection string.
The two failover cluster instances are running in the MSCS cluster consisting of CLUSTERNODEA and CLUSTERNODEB. Other than that, they are completely separate from each other. Each failover cluster instance resides in a different MSCS resource group, and each has a unique set of IP addresses, a distinct network name, and data files that reside on a separate set of shared cluster disks.
When a failover occurs for any resource in an MSCS resource group, all resources that are members of that group also fail over. For SQLCLUSTA, any failure (from the disk resources, IP address, the network name, or the installations of SQL Server within the failover cluster instance) causes all members of the cluster group to fail over when the failover threshold is reached.
The following illustration depicts a two-node cluster with binaries and data. Each failover cluster in this illustration must have exclusive ownership of the disk on which the data and log files are located.