StoreVirtual Storage
1752768 Members
4797 Online
108789 Solutions
New Discussion юеВ

Re: SAN/iQ's reclaim feature using ESXi 4.0

 
Matthew D
Frequent Advisor

SAN/iQ's reclaim feature using ESXi 4.0

I read here (http://www.ostailors.com/docs/servers_storage/thin_provisioning.pdf) that SAN/iQ automatically reclaims zeroed blocks from thin provisioned volumes. We are running vSphere 4.0 with VMFS3 data stores that apparently do not automatically zero-out deleted files. I've read all kinds of crazy ideas for reclaiming unused space but nothing definitive from HP. Does anyone have any experience reclaiming space from a VMFS data store on a thin provisioned Lefthand volume? We have a few VM's that are going to be deleted and before I do so I want to know if it's a good idea to zero-out the virtual disks before actually deleting. From what I could find it looks like Lefthand will reclaim the vast majority of the space taken up by the vmdk file when it's zeroed out but if I just delete it straight from the vSphere client that space won't be reclaimed by the SAN. Thanks for anything anyone has on the subject.
2 REPLIES 2
teledata
Respected Contributor

Re: SAN/iQ's reclaim feature using ESXi 4.0

I think you are misunderstanding the statement:

Here's the quote:
"When you decide to thin provision a volume, it├в s as easy as clicking a radio button in the SAN/iQ administration interface. Thin provisioning helps to keep your volumes thin over time by automatically reclaiming zero-filled blocks and returning them to the free storage pool."

This is true, however what this actually means is if you at first thick-provisioned a volume, and decided to later thin-provision it, it will reclaim any space that the underlying file system has not yet touched (zero-filled blocks). Once a file system (VMFS, NTFS etc.) has used a block, it is NEVER AGAIN considered a "zero-filled block". Since the file systems DO NOT go back and de-allocate blocks when files are deleted, you will always have a minimum used space equal to the "high water mark" or MAXIMUM blocks EVER used by the file system.

Currently the only way to reclaim VMFS datastore "empty" space would be to create a NEW VMFS Volume, migrate the VMs to the new datastore, and delete the old datastore.

Hope this clears things up for you.
http://www.tdonline.com
Matthew D
Frequent Advisor

Re: SAN/iQ's reclaim feature using ESXi 4.0

Thanks for the info, teledata! I guess that just goes to show you can't trust any old blog out there.

Do you know if this will change in SAN/iQ 9.0 + ESXi 4.1 VAAI?