Veeam deduplication and per VM backup files

[andreasaster]

Hi,

we recently upgraded to v9 and are now using per VM backup files. After the first manually launched (the only way to migrate to per VM backup files) full backup of a job, I noticed, that the backup file size increased from 831GB (last full backup with single backup file) to 1,01 TB (last full backup with per VM backup files). Is the Veeam deduplication not working with per VM backup files? Is anyone else experiencing the same?

Thanks
Regards,
Andreas

[Gostev]

Hi, Andreas. Yes, of course deduplication between VMs will not work when you are storing each VM into a separate file. Thanks!

[Josh-VCAP]

Hello,
I see that this is now an option on individual Backup Repositories as well as the new Scale-out Backup Repository. How would this setting effect the total disk space in use by a backup job? Do you still get deduplication across all VM's in a job if the files are stored separately? I'm currently using Reverse Incremental as my backup method.

Thanks

-Josh

[dellock6]

Hi Josh,
the backup method doesn't influence the deduplication ratio, this is managed by the deduplication setting you choose, and the use of single or per-vm chains.

[Gostev]

There are always pros and cons, so there is a couple of important benefits to keep in mind:

1. With Scale-out Backup Repository, while sum of backup file size becomes slightly larger due to the loss of dedupe between VMs, actual storage usage is more efficient because smaller backup files can now occupy extent segments which large multi-VM backup files simply cannot fit. Overall, for scale-out repositories with large amount of extents, this will mean using less storage devices overall despite larger total backup file size. On average, we estimate 10% increase in backup size foot print, but 30% more efficient use of existing storage resources... as you can see, this well covers the loss.

2. Performance! You are basically trading slight increase of total backup files size for incredible performance gain. Early adopters report that moving to per-VM chains and scale-out repository cut overall backup times by 50%! That's a huge win in terms of backup window. And restore performance will also see an improvement.

Finally, keep in mind that you can use per-VM chains for primary backup repositories, and not use this setting for secondary backup repositories where you do long-term archival to. Since our Backup Copy jobs don't just do backup file copy, but rather synthesize new backup files - you can do pretty much anything... for example, you can have all the VMs from multiple jobs each using per-VM chains go into the single target backup file, giving you "global" deduplication for your archive.

[ChrisSnell]

If you want to use per VM backup files and benefit from dedupe across all of the files then you can, of course, use ExaGrid's appliances to achieve this. The 2 technologies work fantastically together and you could benefit even further then from what Veeam has already given you.

https://go.veeam.com/exagrid-and-veeam.html