Defragmentation, sdelete, zeroing virtual disk

[antivir]

Hello.
1.Does defragmentation increase speed of large VM backup?
For example, VM with 500 Gb vmdk, 400 GB free space in windows. Backup takes about 8 hours.

2. Is it safety to use standard windows defragmenter or such program like Rasco Perfect Disk Server?

[Alexey D.]

Hello Andrew,

1. It does - for incremental backup. In case of fragmented disks, a single file may scatter all across the disk, so updating this file makes many block dirty (while incremental changes are picked up with 1MB block, so even if a single bit in the block changes, the whole block is picked up during the incremental pass). And smaller incremental size obviously result in faster backups.

Also, to decrease duration of the backup job, you may create a separate virtual disk for swap file and exclude it from processing in the advanced job settings. This should save you some time, especially when doing the incremental job run.

2. I think any defragmenter would work fine. Be prepared though, that first incremental backup after defragmentation will produce huge incremental file because of many blocks changed. In fact, you may want to perform full backup after defragmentation.

[antivir]

Are there some tools like sdelete but withous problem with free space during it work?

[Vitaliy S.]

Andrew,

You may use any tool for that, but many of our customers are using Sdelete. Also please have a look at the script created by our MVP Even Glemmestad for automated VM disks optimization:
http://www.veeam.com/forums/viewtopic.php?p=2563#p2563

Should help!

[antivir]

There are no that file anymore. Can you share it?
Thank you.

[Vitaliy S.]

Andrew,

I'm afraid I haven't copied it either, but let me try to find it for you.

[antivir]

Thank you. I'm waiting.

[antivir]

One more question. Is backing up (with VSS) during defragmentation safety?

[Vitaliy S.]

Yes, to keep all your VMs with highly transactional applications in a consistent state while doing VM backups/replicas, you should be using Veeam VSS (on all supported Windows machines).

I haven't heard any issues reported on defragmentation procedure and VSS working on the same time. Apparently there won't be ongoing transactions while snapshot is being created. But as Alexey correctly noted I would recommend doing full backup after VM disk defragmentation takes place, in order to have smaller full backup and incremental file sizes.

[Davd]

Here you go Andrew... Even Glemmestads script for defrag and sdelete... http://it.fk.no/div/vm-defragscript.zip

[antivir]

Here you go Andrew... Even Glemmestads script for defrag and sdelete... http://it.fk.no/div/vm-defragscript.zip

Thanks a lot! I will try it tomorrow.

[Shogan]

Thanks for reposting the defrag & sdelete script - was trawling around old posts trying to find it through broken links, till I happened on this thread

[cparker4486]

Hi everyone,

I have a few questions regarding the practice of running defrag and sdelete -z before an active full job to optimize the disks of my VMs. I'm on latest vSphere 5.0 and Veeam 6.5 and am currently testing v7. I performed this process once before but I did it manually which involved making a list and repeatedly checking on the machines to see if their operation had finished before moving on to the next. First I ran defrag then I ran sdelete -z.

My questions are:

1. 1. The reason I intend to do this is because I know it to be the "best" way to reduce the size of a VBK after an active full. Is this true? If not, what is a better alternative?
   2. Does anyone have a Powershell script or other method for automating this process?
   3. Considering that sdelete -z fills up an entire disk is there risk of causing complication in the OS and/or applications due to lack of available disk space? Or is this a safe operation from an application standpoint?
   4. Is there another tool that performs the sdelete task better than sdelete itself?

Thanks!

[foggy]

Chris, please review the topic above for some answers. Here is another useful topic regarding similar question.

I remember other users reported that usage of an automated script that runs before each full backup has significantly reduced the total backup sizes and times. Although you should use this carefully on thin-provisioned disks (as they will be inflated to the fully allocated size), however the disk space available to the guest remains free so there should not be any issues from an application standpoint.