What server performs the compression?

[ssoliman]

Hello,
We have two sites connected via a 1.5MBps PTP with WAN Accellerators at both sites. We have a vCenter server at the main site and a single ESXi server with Veeam BR virtual server at the DR site. We have a Network backup configured to run nightly. The backup runs properly and the WAN Accellerators work like a champ. What seems wiered is the WAN Accellerators show that about twice as much backup data is being pushed over the line then what the actual size of the backup is on disk. Our compression percentage averages about 50% which leads me to believe the compression is happening after the data gets to the DR site (which would be on the backup server). Can anyone confirm this?

If thats the case, is their someway I can change this so the compression happens at the main site (before it goes over the WAN)?

Thanks!
-Steven

[Alexey D.]

Hello Steven,

Yes, compression is happening at the DR site and is handled by B&R server. If you want to have compression at the main site, you should put B&R server there. What is your backup target, by the way?

Please refer to this link for details: Veeam Design Recommendations

[ssoliman]

our target backup is a local drive on the DR site veeam server.

What about dedup, is that happening at the DR site as well?

Thanks!
-Steven

[Alexey D.]

For more information about deduplication, please refer to this F.A.Q. section: Deduplication

Do you have any specific reasons to backup to a local drive (inside VMDK)? It's not considered as best practice - please refer to this post for explanation.

[ssoliman]

Mostly due to cost. That way we are able to use this server as our Veeam backup server and double as a disaster node to power up some of our VM's if we lost our main site.

[J1mbo]

In my testing I found the B&R server needed to be at the target end, otherwise jobs would periodically fail. But then this implies the WAN is running uncompressed it seems.

It seems like an issue development should be looking at (the failures, if nothing else)?

[Alexey D.]

James, which kind of failures do you see, could you elaborate, please? At which circumstances?

[J1mbo]

See ticket 5105901 - with the B&R server at the source end, I always ran into this NFC error after some number of (successful) replications:

Replicating object '[Datastore] vm-name/vm-name.vmdk' CopyLocal failed Client error: There is no FIB [vm-name.vmdk] in the specified restore point.”.

Many thanks!

[Gostev]

This error has nothing to deal with compression or server placement. In fact, this issue is documented as resolved in the release notes for version 5.0.1...

[J1mbo]

Nope, still present in current build and is not seen with the B&R server at the remote site.