Backup Processing Rate

[NerfHerder]

Hi All,

I understand answers may vary greatly, but I am curious what kind of processing rate you see on your backup jobs? And do you see differences in processing rate between virtual machines, physicals, agent jobs, etc.?

Thanks!
~D

[Andreas Neufert]

I saw everything in the field starting from 5MB/s up to a single disk backup with 3GB/s. Up to over 8GB/s with multiple VMs in the Job. Maybe even higher as our software is not the limiting factor.
It really depends on what your source storage is and of cause your target. Then as well how you sice your Veeam components and network.

A single Job with 20+ VMs should run even with incremental processing in a way to saturate 1-2 10GbE links.

[NerfHerder]

Hi Andreas,
Thanks for the reply; Of course I would expect that there is a wide range here and is dependent on the characteristics of the environment and setup. I suppose I could qualify the question a little more by saying that we are a small / mid-size business with roughly 60 VMs and a mix of SAN storage (EqualLogic, Complellent). SAN is on 10 GbE network, but everything else (including VBR) is on 1 GbE.
So, in many ways, an entry level setup. We are just getting started, so...
Thanks!
~D

[Andreas Neufert]

As you have the VBR Server on 1GbE and the actual storage looks good, you should be able to saturate your 1GbE link with the backup. I suggest to use HotAdd mode if you have VMware to doulbe down on the network speed as the data get´s compressed before it goes over the 1GbE link.

[NerfHerder]

Andreas,
When you say "use HotAdd"... is that referring to virtual appliance mode, yes? We have not set this up yet... we have a physical server that is running VBR and has the proxy role as well... So for this, we deploy a dedicated VM which will take over as the proxy / HotAdd virtual appliance, yes? This: https://helpcenter.veeam.com/archive/ba ... iance.html
Thanks!
~D

[Andreas Neufert]

Correct. It can help to boost backup and restore speed 2x potentially (if your 1GbE network is the bottleneck). The idea is the VM will read the data in Hotadd (Virtual Appliance) mode, compress it (reduce the data to less than 50%) and then send it to the backup server (Repository). Same way for restore.

[NerfHerder]

Thanks. For pretty much all backups we are doing so far, the bottleneck shows as "source". Proxy, Network, and Target are pretty much always in the single digits when looking at job load %...

[Andreas Neufert]

Yes, your Source can not deliver fast enough as the 1GbE link which is in this case part of the source connection is the bootleneck likely.

[NerfHerder]

So the next step should be to deploy a VM for hotadd either way... and see how that affects performance...?
Thanks!

[Andreas Neufert]

Correct. For the test, select this new proxy in the job.