Slow Backup - SAN Mode

[KaiseAl1]

Hi,

we see very slow backup speed on full backups of a P2Ved Windows 2003 Server x86 with about 1TB net data. Disks are fresh defragmented and zeroed.
Backup slows down to 7MB/s

SAN is a HP EV4100 4Gbit FC (normal vcb fullbackups with about 120MB/s to a fast raid 0 destination)
Veeamserver is a older HP DL385 with 4 cores and 4 GB RAM
Backup destination is a HP MSA20 SATA enclosure directly attached to the Veeam server (seq write speed about 50MB/sec)

ESXi version 4.0U2
Veeam version 4.1.2

Any ideas to speed up our backup?

thanks and best regards

Alex

[Vitaliy S.]

Hello Alex,

Please look through the topic below and please try to answer Anton's questions (from 1 to 3), it will be helpful for futher troubleshooting:
http://www.veeam.com/forums/viewtopic.p ... ode#p19552

By the way, could you tell me whether you do observe the same slow speed rates for other VMs or not.

Thanks!

[KaiseAl1]

Hi Vitaliy,

thank you for response. I think I got a little bit closer to the problem.

First, the problem occures on this VM only. But:

Because it's a large VM (4 disks: system 18GB, data 200, 600 and 800GB) I started the first full backup after Job creation with the 800GB disk excluded. The performance was OK (~40GB/s)
During the following night, I added the last disk (800GB) to the job. I expected that the vbk file would grow and a small vrb file with the incrementals from the other 3 disks would be created.
The result was a heavy growing vrb and a near stable vbk.
Also the performance drops down to about 7MB/sec

Maybe I missunderstood vbk and vbr

Last night I recreated the backup job with all 4 disks included. The job is still running at 37MB/s

best regards

Alex

[Vitaliy S.]

Alex,

Could you tell me please if you have any other P2V VMs in your environment? I'm asking this, because I do remember another topic with similar issues:
http://www.veeam.com/forums/viewtopic.p ... eed#p16767

On top of that, I'd say it is expected to have different backup speeds for different VMs or VMs with new disks added. But your VBK/VRB understanding is correct, though VRB file growth is also expected as we need to store differential information between VBK and VRB files in there.

If I were you I would check the deduplication and compression ratio for your VBK file (you need to navigate to Backups and choose Properties to get the information), this could shed some light on the reason why you have almost the same size for the VBK file.

[KaiseAl1]

Hi Vitaliy,

it's the only P2V VM in our environment.
The first full-backup with all disks selected finished yesterday with 37MB/s average speed. This seems correct to me.

I think, the trouble started only with my approach to backup the VM in stages. But it's still a mystery to me why the vrb was growing that much.

Compression ist "Optimal" an Dedup is "On"

best regards

Alex

[Vitaliy S.]

Alex,

I was referring to the statistic about actual numbers for dedupe and compression for the VBK file, so could you please check that. Besides, what is the vrb size if you trigger a full run and then an incremental pass with 800 GB virtual disk excluded? Is it the same as you had with 800 GB not excluded from the job?