Hi,
Is it still recommended to defrag virtual hard disks using vmdk files and vmware 5?
I'm having VSS backup issues with a 2008 R2 server and was wondering if defragging the hard disk would make any difference. My Exchange 2010 Server gives backup errors with VSS.
Thanks
VMware and Storage Vendors recommend Don’t defrag VMs!, refer http://www.ntpro.nl/blog/uploads/Mythbusters_Dutch_VMUG.pdf
VMware and Storage Vendors recommend Don’t defrag VMs!, refer http://www.ntpro.nl/blog/uploads/Mythbusters_Dutch_VMUG.pdf
What are the VSS errors? Could it be associates with some other automated task or perhaps a latency issue with snapshots? What backup utility are you using to pull from the VSS?
VM aside, defrag on any hardware with RAID doesn't make sense.
Surely your VM environment utilises RAID.
Our SAN utilises RAID.
The VSS errors are failure querying VSS writer - which is an Exchange 2010 error when using Symantec backup exec.
To be honest I didn't think defrag was the answer, but thought the question was worth asking
Josh26 wrote:
VM aside, defrag on any hardware with RAID doesn't make sense.
This is a very interesting topic I think. Could you expand some on this opinion?
In my mind it could make a lot of difference with a fragmented disk even on RAID - if using Windows default cluster size of 4 KB and you should read a, say 32 KB, file which is totally fragmented. This would make the Windows OS send 8 read-request at different places at the disk, which on the RAID system most likely will have to be completed on different disk spindles. This would make this single file read cause 8 IOs against several disks in the RAID group.
If the disk was defragmented and the 32 KB file was stored together a single disk IO could be sent from the Windows OS and there is a high chance that this would be allocated inside one "stripe size" on the RAID group, generating a single IO to a single disk.
This is just some thoughts - I do not know if there is any definitive "answer" to this.