When deploying SQL Server FCI with WSFC, consider the storage layout. SQL Server database and transaction log files use a file system directory structure. When using external storage such as PowerMax for the SQL Server databases, each storage volume can be configured in Windows Disk Management as a drive letter or a mount-point, or in Failover Cluster Manager as a CSV:
- Drive letters—Drive letters are the simplest way to identify the location of SQL Server database files. There are only 26 drive letters, which limit the overall scale, especially if multiple database copies are mounted to a single host.
- Mount-points—When presenting disks in Windows Disk Management, a mount-point can be selected rather than a drive letter. Mount-points are subfolders (subdirectories) of a root mount-point (root directory). The root mount-point must be on shared storage and have a drive letter. When a root mount-point is determined, it can have many subfolders on it, supporting high-scale. Each such subfolder can have its own storage volume behind it. This option is both scalable and flexible.
- Cluster Shared Volumes (CSV)—To reduce the number of times roles move from one node to another in WSFC, select CSV. Unlike the previous two options that can be configured in Windows Disk Management, CSV is configured in Failover Cluster Manager by designating available disks as CSV. When a disk is designated as CSV, it is available to all cluster nodes and managed by WSFC. It is no longer a resource that needs to failover between nodes, making such events faster. The drawback is that all CSV disks appear under a single root directory on the C:/ClusterStorage/ system drive. For more information about CSV, see Use Cluster Shared Volumes in a failover cluster.
Due to the advantages of failover speed, it is recommended to use CSV. For better flexibility and scale, mount points are recommended. For small deployments, use drive letters.