fragmentation

[pfeifferl]

Good question! I'm also interested in that ... - open-e, what's best practise?

BTW - Another question to you, Jon: Is your Windows 2008 R2 cluster with DSS v6 running smoothly?? No troubles with BSOD/reboot of VMs while autofailover on DSS side? Or do you not using DSS failover?

[enealDC]

Keep in mind that DSS does not understand the underlying file system on the LUN. With DSS, the disk is carved into Logical Volumes which are served to your servers. With that said, no tools are provided to do any defragmentation. All of that needs to be performed on either the hostOS or the guests themselves.

Here are some tips


1. Preallocate space 100% of the time for any VS
2. Try not to host different file system types on same lun
3. Try to use different luns for swap OR don't use swap at all
4. Try not to use different block sizes on the same lun
5. Use different luns targets for the OS and the data (where possible)
6. Use NFS were possible to host application data away from the OS lun

[JonMoore87]

Hi Pfeifferl,

I'm not having any problems with BSOD or reboots of VMs, but as you suggested, I'm not using auto failover on the DSS.
It is something that I'm interested in but from what I've read on the forums it doesn't seem to be at a point where it's ready for hyper-v.
So far I've found open-e to be more VMware biased but hyper-v seems to be coming along.

Having looked further into defraging the CSV, I've found if the host machine were to perform it you'd have to switch on redirected access so that all the files can be locked by the one host performing the defrag. Pretty bad for performance.

I was hoping for some sort of continual job to run but it seems that this can't be done at the moment... or not without having all hosts access the storage via the host that owns the CSV.

For now I'm just going to make sure I keep my guests defraged.

Keep me informed of your progress with failover, aside from the additional DSS it sounds like our setups are pretty much the same.

Cheers,
Jon