Really slow read speeds: DD160 & Veeam

[Lord-Data]

Hi guys,
I've got a setup with Veeam doing snapshot backups to the DD160, no problems at all there, runs quite fast, and works well. I then do a weekly Copy -> Tape job, for an off site copy.

The DD160 is mapped as a NFS share for both read and write operations, however it should be noted I've done this via CIFS as well, and no difference. (NFS is faster for writes, but no difference in the read/copy to tape speed)

When I say speed issues, I mean it is running right now at 12mbps, where it writes to it at up to 100mbps. This is a job with a total of 2.5TB, with veeam storing that in 2 files (master and one incremental)

I'm curious, is the size of the files causing these issues? I've got another job (SQL direct backup) that puts SQL backups to this same DD160, via CIFS, and then copies to tape daily, and this does Files -> Tape (same server, same DD, same Tape drive, same everything) at > 70mbps, so it's not the physical devices or interfaces causing the issues. If I split to job up, into say 10 separate VM -> Disk jobs, I would end up with 10 separate files on the DD, each much smaller, would this run better?

Any other ideas on where to look as to why this one is so slow, but other file copy jobs from the same source work better?

[jpaul]

Can you take a screen shot of the stats screen on the DD160 during the tape copy job ? I find that when i do a tape copy from my DD160 in my lab im getting 50-60MB/s (in line with your sql job)... So it might be a little hard to say why the other is dragging without seeing some stats.

Thanks!

[Lord-Data]

Ok,
http://imgur.com/a/iLHBW

Current stats. This job is now at 48%, after running for 2 days. Not going to work out. This is the worst it's been yet. I'm going to cancel it (so I can put my SQL -> Tape job down, its a bit higher priority), and then try a new job of a single VM to DD, and then copy that to tape, and see how it performs. I'm really thinking it's related to the large single file, versus much smaller files.

[jpaul]

well judging by that the DD shouldnt be the issue even on this job.

What is the RAM usage like on your Veeam server that is involved ? only other thing im thinking that could be different is that the metadata (dedupe lookup table) for a 2.5TB backup would be much larger than one for a smaller backup....

Other than that thought Id say give Support a call... they are pretty good!