Veeam max speeds?

[thomasrensvold]

Hello!

I'm using veeam backup and replication v7 with patch 1. Using this for backups of 100 VMs on vmware. (vsphere 5.1 u1).
Veeam server is a virtual machine with 16 cores and 24GB memory. It's running on windows server 2012 now. I also tried 2008 R2.

I cannot get better speeds than around 110 MB/s (reported by veeam). I tried experiencing with different number of concurrent jobs/tasks and also several proxy servers, but I can't get any higher total speeds. Here is a screenshot of freenas network monitor, seems like it's capped on 600mbit:

https://www.dropbox.com/s/6xs1u3ynyl223 ... .55.51.png

The peak you see above 600mbit is another backup job, not veeam.

See example from veeam console here with 2 proxy servers and 15 concurrents tasks in 1 active full job:

https://www.dropbox.com/s/j6b3630tfd4v3k8/29.10.jpg

Bottleneck says

Source 0%
Proxy 1%
Network 0%
Target 13%

So there's not really a "bottleneck" here.
My backup repository is a freenas server. When I do a regular windows CIFS copy (from the veeam server to the backup repository I get a lot better speeds than with veeam backup, I get at least 3gbit when I do tests.

Both servers is on 10gbit ethernet. The freenas server is a physical server.

I tried both CIFS and iSCSI as backup repository.

I have enabled jumbo frames on both servers and verified that it's working by pinging with larger packets. Does not help.

Is there any limits on veeam causing this? What could I do to get better speeds?

[foggy]

Thomas, you cannot compare backup job performance with the performance of simple file copy, since backup is not just about copying a file from Veeam server to repository.

Bottleneck stats shows that target storage performance is primary bottleneck, so I suggest you to test backup to some other storage (Veeam backup server local disk, for example) and see the performance rates. Also, what are the bottleneck stats for the entire job (those you've reported seem to be the real-time stats during the job execution).

[Gostev]

Bottleneck says

Source 0%
Proxy 1%
Network 0%
Target 13%

So there's not really a "bottleneck" here.

These stats are incorrect, as at least one should be close to 100%.
Please let me know your support case ID, as this specific problem is interesting.

As far as your real bottleneck, I can guess it is your backup storage.

[yizhar]

Hi.

Can you also try the following tests:

Configure a physical host as repository - run some backups against it.

Configure a physical host as proxy and let it handle some load then compare results.

Yizhar