Discussions related to exporting backups to tape and backing up directly to tape.
Post Reply
vmtech123
Veeam Legend
Posts: 235
Liked: 133 times
Joined: Mar 28, 2019 2:01 pm
Full Name: SP
Contact:

a few questions

Post by vmtech123 »

I am using Veeam 11 and Windows 2019 REFS repos.
I have V7000 SANS where I can easily push the IOPS and throughput.

I just upgraded from lto6 to lto8 and find my speeds vary quite a bit. Sometimes I can hit 337MB on a single drive, but at others it is 120MB~ across the board making for very long backup windows.

At my Main site, I backup using forward incremental, along with a few reverse incremental.
I run a Copy job to the DR site keeping a few weekly's and monthly jobs.. From the DR site I am doing backup to Tape using GFS. (I am setting up tape at the main site soon too)

Should I be doing active full backups to get more speed out of my Tape backups? Is there an issue with using a Copy Job \ GFS then a GFS Tape backup?

I just wanted to keep weekly jobs, do I just create NON GFS jobs and uncheck the incremental tab?

I'd prefer to have weekly and monthly jobs, Keeping the monthly jobs for a longer retention. Perhaps even a yearly backup. If I want to avoid the daily backups, do I create 3 separate NON_GFS jobs and remove incremental?

I'd first like to get the throughput higher. The Tape drives have hit great speeds with Veeam, so I am wondering if it is rehydrating, or if there is fragmentation causing it.

The Target seems to show up as the bottle neck at 52%, Network 18, proxy 3, source 0
vmtech123
Veeam Legend
Posts: 235
Liked: 133 times
Joined: Mar 28, 2019 2:01 pm
Full Name: SP
Contact:

Re: a few questions

Post by vmtech123 »

Interesting enough, I tried using 2,3, and 4 drives at once, and the speed didn't seem to increase much after 2 simultaneous.
1 tape is about 340MB
2 tapes is about 500-640
3 tapes or 4 tapes was about 500-600

I tried it from the Main SAN over our 10gig link and got roughly the same throughput to the library at the DR site.
At the DR site the proxy is direct connected to the SAN as the REPO with a large volume. 16GB fiber to Proxy/repo. (It is also currently the TAPE server while I do some testing - If the speeds can increase I'll get a standalone TAPE server)

You would think direct from SAN to TAPE should be able to push these drives.
vmtech123
Veeam Legend
Posts: 235
Liked: 133 times
Joined: Mar 28, 2019 2:01 pm
Full Name: SP
Contact:

Re: a few questions

Post by vmtech123 »

One more test
For 3 jobs I did active full backups.

1 Drive 340MB
2 drives - 680MB
3 drives I got up to 1GB for a bit.

I am guessing that even though my V7000 has 144 disks at the DR site, trying to rehydrate or create the synthetic backup is just taking too long?

Active Full backups are not really going to work as I have many Servers in the 20TB+ range an it's going to require quite a bit of backup space. At least I might be on to a better solution now seeing the speed increase after an active full.
Dima P.
Product Manager
Posts: 14396
Liked: 1568 times
Joined: Feb 04, 2013 2:07 pm
Full Name: Dmitry Popov
Location: Prague
Contact:

Re: a few questions

Post by Dima P. »

Hello SP,
Should I be doing active full backups to get more speed out of my Tape backups?
Tape jobs with active full backups are faster due as we do not need to build the synthetic fulls. Last fully depends on the storage performance and the way how data is stored, say if it's deduplicated we need to rehydrate the blocks prior streaming those to tape.
Is there an issue with using a Copy Job \ GFS then a GFS Tape backup?
Nope.
I just wanted to keep weekly jobs, do I just create NON GFS jobs and uncheck the incremental tab?
Can use regular backup to tape or GFS one, it's up to you. In terms of performance they are identical.
I'd prefer to have weekly and monthly jobs, Keeping the monthly jobs for a longer retention. Perhaps even a yearly backup. If I want to avoid the daily backups, do I create 3 separate NON_GFS jobs and remove incremental?
In such case GFS media pool is better - removes the need to monitor three jobs and extra fulls (with GFS jobs whenever full backup needs to be created for multiple media sets - it goes only to elder media set).
I'd first like to get the throughput higher. The Tape drives have hit great speeds with Veeam, so I am wondering if it is rehydrating, or if there is fragmentation causing it.
Is it possible to run a test and tape out the vbk from non ReFS partition to see the difference?
1 tape is about 340MB
2 tapes is about 500-640
3 tapes or 4 tapes was about 500-600
If you are talking about parallel processing - you need to have multiple backup chains at source to see the performance boost. Say, you have tree vbk files - 4 drives wont be faster than three drives.

Thanks!
vmtech123
Veeam Legend
Posts: 235
Liked: 133 times
Joined: Mar 28, 2019 2:01 pm
Full Name: SP
Contact:

Re: a few questions

Post by vmtech123 »

Thanks for the reply. I didn't realize I could just uncheck Daily backups from the GFS and keep it as weekly / monthly etc. This is much better plus control over appending the jobs will save me a ton of tapes, or exporting my monthly's.

I increased the Queue Depth on the Veeam server and it seemed to help. At the main site using 4 drives I am getting 1.2GB to 1.3GB so that is between 300GB and 325GB each. Not too bad. If I run a single drive I'll hit 360 and with two I can hit 700 so I still think there is some queueing going on but I'll dig into that later and do some testing.

On 6 Drives at the DR site from the GFS backups I am getting 1.3-1.9GB... When I did some active FULL backups I was able to get over 2GB using 2 drives, so the fragmentation slowing me down I think.

Unfortunately we have lots of 20-35TB servers so doing active full backups are never easy to schedule. Things tend to fall behind as they take too long. Maybe I need to ask for some flash storage for our backups :)

If I continue to use the GFS backups, does the Active Full replicate the full backup or just an incremental to the second site? If it's the latter it wouldn't help much on the second library but it would on the main site.


Also, a feature request I have would be a scheduled tape copy job, or something along those lines. If I have a Media Pool at one site that I want to do a file backup to, I want to have it replicate to a second site. That would save me having to configure 2 of everything and have redundancy. An extra feature would be to restore the media pool or media set if a Tape breaks. It would only have to copy back one tape and verify it's all good.
Post Reply

Who is online

Users browsing this forum: mrmccoy007 and 21 guests