What dedupe are you seeing for Veeam Agent 2.0?

[mtellier]

We have a Windows 2012 file server that is a VM hosted on VMware. The file server has iSCSI LUNS direct to Nimble Storage Volumes and have installed the Veeam Agent for Windows version 2.0 for backup.
The backup agent is working, but were getting de-duplication of 1.2x on Volume level backup. Our test backup consists of 2 volumes and 1.1TB of files, like user home directories and department shares. Were planning on doing a weekly full backup and daily incremental, this one server hosts about 6TB of data to backup. We typically see > 2.x for block level backups in a normal backup job.

Here is our configuration (Full Backup):
Compression Level: Optimal
Storage Optimization: LAN Target
Processed: 1.1TB
Read: 1.1TB
Transferred: 964.6GB (1.2x)

Backup storage is a Synology Rackstation with 10Gbps interface.

I contacted technical support, they had me change Compression to Dedupe-friendly and Optimization to WAN target. Using these settings there was absolutely no savings on the backup size, Processed 1.1TB, Read 1.1TB and Transferred 1.1TB.

Anyone else experience the same de-dupelication for file backup? What about Exchange backup with the agent?

[Dima P.]

Hi Mark,

Good deduplication results can be achieved when you are using a deduplication storage. For instance take a look at this post - it shows impressive savings while using native Windows 2012 deduplication repository.

[SDMatt]

I'm seeing almost zero deduplication with Veeam Agent 2.0, at least when using the file level backup option.

To test this I copied a 600 MB folder to the machine, and then created a second identical copy of that 600 MB folder. My backup grew by 1.17 GB.

I'm working through a volume level backup as a test now.

[SDMatt]

I used the same method that I did in the previous post for a volume level backup... and I see the same results. Zero noticeable space savings.

To test I created a volume level backup.. ran the 1st full backup, then ran another to establish a minimum baseline for the file size of an incremental. I then copied a 600 MB folder to the machine, and then created a second identical copy of that 600 MB folder. Then I ran a 3rd backup. If dedupe was really working I'd expect my backup would increase in size by 600 MB, but it increased by about 1.2 GB (Settings used Extreme Compression / Wan Target).

For giggles, I changed the settings to Compression to Dedupe-friendly and Optimization to WAN target and re-ran the test.. backup size was 1.3 GB

IMHO - this product doesn't dedupe.

[SDMatt]

Page 2 of the Editions Comparison document lists built-in deduplication as a feature of the free version of the product.
https://docs.google.com/document/d/1EWB ... sp=sharing

Here's a link to the doc
https://www.veeam.com/veeam_agent_windo ... son_ds.pdf

Something isn't right here.

[mtellier]

After adding additional volumes totaling 7.1TB, a full backup size is 5.2TB or 1.4x backing up to a NAS.

[Dima P.]

Matt,

Deduplication engine was taken from our flagship product Veeam B&R. In Veeam B&R we see the best deduplication ratio when VMs with similar operating systems are being backed up (frankly speaking, the best thing to dedupe is operating system files). If you want to achieve good deduplication ratio between you backup files – try to use repository with deduplication feature, for example Windows repository with deduplication feature enabled.

[SDMatt]

Thanks Dima,

Veeam's inability to perform decent deduplication without some sort of add on (3rd party software, deduplication storage device) has always been on my list of beefs with the product.

I'm disappointed to find out that this is working as designed.

[Luca82]

Hi everyone, is the deduplication done among several machines? So if I have machine with the same OS/Data can I expect to have an high deduplication ratio?

[Dima P.]

is the deduplication done among several machines?

No, deduplication is per job. Resulting backup files are independent from each other, so it’s recommended to use a repository with dedpucation engine on board (for instance Windows Server with deduplication role).

[Luca82]

Thank you. Surely deduplication is per job... but if I make a single job with several machines ....deduplication is not among machines so.....deduplication is or not per job? Sorry but I don't understand.....

[Dima P.]

It's impossible to create an agent job with multiple computers since agent is installed per computer and Veeam B&R’s backup copy job does not have built-in deduplication.

[Frenchyaz]

I got the same thing. Seems like VB&R uses zeroed thin deduplication...
Now we are doing deduplication straight on the repository using Windows deduplication. A 10.8TB data is 2TB on disk after dedup... not too shabby.

[Dima P.]

Good result, thanks for sharing! To increase the dedulpication ratio you may want to disable the backup compression (under the advanced settings on the repository step). Cheers!