Veeam, DataDomain and Linux NFS share

[zak2011]

Is there any guide on how to configure the NFS share on the Data Domain with a Linux VM?
The Linux VM i am using is Ubuntu.

Thanks

[foggy]

Arun, just mount the storage directly a Linux VM and add this VM as a backup repository to Veeam B&R console.

[zak2011]

I tried mounting the directly on the Linux VM. However still not being able to add the Linux repo in Veeam.
following instructions from

http://www.cyberciti.biz/tips/ubuntu-li ... share.html

[foggy]

However still not being able to add the Linux repo in Veeam.

Can you access the shared directory after mounting it? What issues do you encounter when adding repository in Veeam B&R?

[zak2011]

On the Linux VM it says it has suceesfully mounted. However when I og back to Veeam to add the repository. After the stage that I put in the IP of the Data Domain, i need to specify the path to the folder.
So i give /mnt/newDD and if I try to populate it says 'Unable to detect Storage parameters for path /mnt/newDD'.
I have given full access to the Linux VM and the Veeam server on the Data Domain NFS share .

Thanks!

[tsightler]

After the stage that I put in the IP of the Data Domain, i need to specify the path to the folder.

Perhaps this is just a typo, but you don't give the IP of the Data Domain, but of the Linux server. The Linux system is effectively acting as a gateway device to the DD.

[tomas.olsen]

that might be a good idea. if it was possible to use disk as primary target and keep the seven first restore points to disk. this would give you fast restore times, surebackup and everything. And then stage let say the three, seven or eleven sequential weeks to datadomain (i addition to the first week on disk).

[zak2011]

Absolutely Tom. You got it right. That was the key. I simply didnt realize that I should have added the IP of the Linux server and not of the Data Domain.
Thankyou so much Tom for enlightening me on this. It was a really silly mistake on my part.

Thanks again!

[chrisBrindley]

I have been on the forums most of the day trying to solve the issues we are seeing with restores in our environment.
Nothing makes me more frustrated when a customer calls in for a file to be restored and i am sitting here for 9:38 for the backup browser window to show the files, when we tried a total VM restore it takes nearly an hour.

I have followed all the guidelines possible in the forums, we have proxys using hot add, we run our jobs with dedupe friendly compression and LAN target setting we do all this to a DD860 on a 10g network to a CIFS share.
backup speeds are over 70 - 100MB
I can copy files manually to and from the DD860 share in windows and get up to 70MB transfer rates either way.

So where do i look now, do i open a ticket, will any logs show any issues

[foggy]

Chris, slow DataDomain performance on backup browsing/restore is actually a known issue. You can find some details in the thread above, as well as a couple of hints to speed it up (using the Other OS Restore wizard and running defrag prior to backup are among them). Thanks.

[tsnhelpdesk]

So I am having an issue with the DD2500 and Veeam 7 restores.

My backups to the DD2500 are blazing fast however my restores are painfully slow.

I have the Veeam compression and dedupe turned off on my backup jobs.

My backup jobs process at over 100Mb/s and my restore for 100GBs is running at 13Mb/s.

Is there something I can do to speed this up?

[veremin]

The speed of restore operations from dedupe storage might be a challenge, indeed. Please, see answers provided above for information about potential tweaks that can help to increase performance. Thanks.

[Gostev]

One new recommendation we have worked out with Veeam users leveraging deduplication storage recently, is using 8MB block size for Veeam backup files. This can be enabled by enabling "Local target (16TB+ backup files)" storage optimization under advanced backup job setting. While incremental backup may become a little slower, this does not seem to impact backup footprint due to deduplication in storage, while the restore performance from backup files created with this setting is significantly improved. Hope this helps, and it would be great if you could update us with the new results! Thanks!

[matai]

We've been backing up directly to a Data Domain using NFS and a Linux proxy. I'm working on getting a landing zone in between and also the BOOST license for Veeam 8.

Either way, I just created a CIFS share for a separate purpose. Will there be any performance difference between sending the Veeam backup jobs through the Linux proxy to the NFS share versus directly to the CIFS share?

Thanks!

[matai]

Not sure if this helps or causes more confusion, but our current processing rate on NFS is 2-3MB/s doing reverse incrementals. All our VMs are on a Nutanix cluster so we have the transport mode set to network.

These are some of our bottlenecks:

10/15/2014 10:58:41 PM :: Load: Source 13% > Proxy 0% > Network 0% > Target 97%
10/15/2014 7:55:45 PM :: Load: Source 10% > Proxy 9% > Network 0% > Target 88%
10/15/2014 9:24:40 PM :: Load: Source 3% > Proxy 0% > Network 0% > Target 84%

Thanks again

[foggy]

Please review the tread above for some considerations regarding your question. There is also another existing topic that could be helpful.