Slow disk group failover when using Microsoft Volume Shadow Copy Services

Problem

If a volume in a Cluster Disk Group is configured to use a different volume in the same Disk Group as a destination volume for Volume Shadow Copy snapshots; then slow Disk Group failovers can occur in a clustered environment.

Cause

If a volume which is a destination volume for Volume ShadowCopy Services (VSS) snapshots is taken offline prior to the host volume then the shadow copies are aborted prior to the dismount.  This can cause open handles to delay the deport operation and VSS cleanup to cause delays of the import operation.

Solution

If using Storage Foundation HA for Windows, dismounting of volumes is controlled by MountV resources.  Create a dependency between the data volume and the destination volume so that the data volume is taken offline first.

In a Windows Failover Cluster (WFoC) environment, it is not possible to control the volume dismount order.  In this case, it is recommended to abort all the current snapshots and create an additional Cluster Disk Group containing the VSS volume.  It is then possible to create dependencies using WFoC features.

Note: It is not allowed to split a Disk Group per volume when there is a diffarea for another volume defined on that volume, Storage Foundation for Windows will block the operation.

Reference: http://support.microsoft.com/kb/838421