Slow Cluster Shared Volume–Tips to Speed Up CSV and Backups

After the introduction of update 2878635(http://support.microsoft.com/kb/2878635) and update 2903939 (http://support.microsoft.com/kb/2903939), you now have a CSV Software snapshot life cycle management mechanism to fine tune VolSnap and VSS when using CSVs.

It’s not uncommon for snapshots to add up and not be cleared by defective backup applications. Some backup applications are designed to use persistent VSS snapshots.

When Windows has to manage multiple large, and fragmented VSS snapshots, each drive access becomes much slower than when no snapshots are presents. There’s a good chance the delays you are seeing are due to having too many VSS snapshots. Fragmentation is also common.

Microsoft offers now two new settings: SnapshotDiffSize and SnapshotAgeLimit

 

You can use these two new settings to fine tune ‘snapshot death’ or automatic renewal when a size or age limit has been reached. See this article:

http://support.microsoft.com/kb/2935810/en-us?sd=rss&spid=16526

However “trashing” VSS snapshots, unless they are orphans, may be an issue if you are using backup applications that require persistent shadows.

BackupChain does not use persistent backups and you should never see orphaned shadows when using BackupChain.

Windows Server Backup, on the other hand, and a couple of others are known to use persistent shadows which remain even after reboots. To delete those shadows (at your own risk) use VSSUIRUN.EXE or VSSADMIN from the command line.

System Restore is another feature in Windows that may auto create shadows which eventually clog the entire system. Those shadows are usually only taken of the system drive, not cluster shared volumes.