Prevent Veeam from backing up white space?

[marine]

I have already opened a support ticket on this (ID#5170617) but have not yet received an answer. Maybe someone here can help me? I am using Veeam v5 and ESXi v4.1 but I recently discovered a nasty issue. If I copy a large amount of data onto my file server vm and then erase it again before my next backup... this still results in a large incremental .vib file. Even though the size of the vm has not changed. I tested this by copying 20GB of data to my file server, then deleting it, then performing an incremental backup and it resulted in a 20GB .vib file! Surely there must be a way for Veeam to recognize that this is white space and no longer used?

These are vmdk's that have been deployed in thick format. For my tests, the total size of my vm is 65GB but only 21GB is used. The other strange thing is that if I erase all of my test backup jobs and start from scratch, the first full backup results in a 48GB .vbk file but the actual data used in the vm is only 21GB. Why is Veeam backing up this empty white space?

I am not sure if this is a limitation of Veeam or ESXi or if it is just a bug of some type? Either way it is a definite show stopper for us and we will need a new backup solution if this cannot be solved. I have read some older posts on using sdelete to zero out blocks etc but that is not a good solution because I would have to run it each night before my backup job and it would take way too long.

Any help would be appreciated. Thanks.

[foggy]

Actually, this is neither a limitation/bug/whatever of Veeam nor VMware but a "feature" of Microsoft. This is just how NTFS works so nothing can be solved here from the Veeam side.

[marine]

WOW! That amazes me! Talk about a show stopper!

This really sucks because Veeam works really well in so many ways. But it looks like I will have to go back to using some sort of agent based backup solution. Is there no way that Veeam can access the Master File Table for an NTFS file system and determine which blocks are white space? Maybe a Veeam Agent is in order, kind of like the VMWare Tools that get installed inside each vm? You could have a Veeam Tools Agent that is optional for anyone wanting features like this.

[Vitaliy S.]

If I copy a large amount of data onto my file server vm and then erase it again before my next backup... this still results in a large incremental .vib file.

Out of curiosity...how often do you do this? If it is indeed a regular operation, could you please tell me what data do you copy and erase every time prior to the backup job?

[marine]

We are implementing an archiving solution that automatically moves files from primary SAN to backup SAN and leaves a "stub" in place. So to the end users it looks like their file is right where they left it on their network drive. But we have some research projects that will be bringing back up to 1TB of data from time to time and needing to store it. For some projects this will happen 2 or 3 times each year but for other projects it can be every few weeks. Our plan was to ingest this data through our file server in chunks. Thus, the size of our archive database would grow but the size of our file server vm would remain the same. And the Veeam backup window would not grow.

Seems like I will need to find another solution if this will cause a huge .vib file each time.