Slow restore or how find bottleneck in restore job

[JosueM]

Good day everyone.

Veeam runs a full backup a 2.5 TB mssql server vm in 1 hour 41 minutes trasnfering 413.4GB at 404 MB/s.

Then it restores the same VM in 10 hours 20 minutes at 70 MB/s.

The used infraestructure is the same for both jobs with the following details:

Vsphere:
3x poweredge R710 hosts with 4 dedicated 1Gbps
2x dedicated iscsi switch 1Gbps with 4 vlans
1x dell MD3220i iscsi cabin with dual controllers and each controller with 4x1Gbps for iscsi traffic.

storage volumes both on the MD3220i cabin:
-productive volume 6TB raid5, 9x800GB SSD hard drives
-restore volume 9TB raid6, 18x600GB 15k RPM hard drives.

Veeam:
2x veeam proxy servers with 8 cpu one of them and 6 cpu the other uses hotadd in backup and restore.
1x windows 2012 physical backup server repository with:
- 4x 1Gbps Nics
- 5.5 TB volume storage raid 6
- 1x PERC H730 controller
- 8x 1TB 7.2K RPM SATA hard drives.


Any idea or suggestion to improve the restore time?

Thanks in advance

[DGrinev]

Hi Josue,

Did you check which transport mode was used for the restore operation? Thanks!

[JosueM]

Hello DGrinev,

I guess the transport mode used was virtual appliance since the restore log shows the message:

Restoring Hard disk 7 (750.0 GB) : 677.3 GB restored at 34 MB/s [hotadd]

[foggy]

Any chance dedupe is enabled on the repository?

[JosueM]

hello Foggy,

dedup is not enabled.

[foggy]

Since there's no bottleneck stats for restore jobs, it's pretty hard to guess where the bottleneck is, so I recommend asking support for assistance in reviewing your setup and identifying the slowest link during restore.

[JosueM]

hello Foggy, did that on ticket # 03146866 and tech expert told me the performance was the expected for such scenario and the size of that vm. So i guess there is nothing than can be tweak/tune to improve the job times.

thanks