After Upgrade Backups Appear Slower

[GLaDOS]

After upgrading to 12.5, and finally getting all of the remote agents upgraded, one of my backup jobs appears to have gotton a lot slower. May just be a network issue, but the backups before the upgrade were running at ~310 MB/Min, from last night a backup is still running and it's showing ~30 MB/Min.

Has anyone else noticed a speed issue since the upgrade? If not, it's time for me to look a little bit deeper and see where my bandwidth is going.

[GLaDOS]

After canceling the slow backup job, I noticed that the history ended up reporting that it ran at 270MB/Min.

This is making me a little concerned about speed reportings accuracy in backup exec, and why isn't there a way to get a better speed from backup exec, for example something like multi-threaded backups, and cacheing, instead of network device > backup device, do something like multiple networked devices > a sorted cache > tape.

Secondly, why isn't there a speed graph that is made during a backup? showing speeds, and intersecting points at where it changes to a new device... etc. (Live graphs wouldn't be needed, IMO, but something to do some diagnostics.)

[GLaDOS]

I'm getting really miffed that I haven't gotton any responce here. Where's the customer support?

Granted this issue isn't a critical down issue, but still where's the support?

I'm highly disapointed with the forums here.

I have simple questions, like, how is the "Job Rate" Calculated? is it total time over byte count? is it an avg of recorded speed per connection? 

Why hasn't there been any responces from symantec here?

[Kevin_Cotreau]

I also recently upgraded from 11d to 12.5 and have experienced a similar significant slowdown. My backups on a Quantum SuperLoader 3 LTO-3 drive went from taking just under 8 hours to just under 10 hours.

First Glados, I don't think it has anything to do with your network performance. I am a one-man company, so I only have one server, and everything is backed up locally for me, and I still experienced the slowdown.

I even tried uninstalling and deleting all traces of BE and then reinstalling BE 12.5 with no change. I also tested a few tapes by doing a long erase, and still no change. I am also confident in my hardware since Quantum replaced it less than a month ago when I did have a hardware failure, and it had been working perfectly until the upgrade.

Second Glados, I would be interested to know if you upgraded from 11d, or what version? Also if you have logs from the previous version when it was faster, could you compare them to the current logs? I noticed that my backups were not actually slower, and in some cases, like for Exchange and SQL server, they were actually faster. My slowdowns were all in the verify portion of my backups.

This is not a real problem if they actually have added some additional checking during that verification process that may improve backup validity, but if it is just plain slower, that is a problem.

For now, you can do one of two things: If you find your slowdown is, in fact, due to verification, you can choose to verify less often, like maybe during a weekend backup only. Or, you can choose to change the way you do your jobs. I had a lot of static virtual machines that I use for testing only that I excluded from my daily backups and now I only back them up over the weekend for my off-site tape.

Please post again and let me know if you found your slowness to be during the verification portion of your backups. I am betting it is and that we are not the only ones who will have this issue.

[ITCowboy]

We are having the same problem, except even slower. Our current backup runs about 700 GB (4 Tapes on HP Ultrium II) and runs over a dedicated network. On 11 D, this would take 8-9 hours. Since upgrading to 12.5, the exact same backup, takes 6 hours for 1 tape, the whole job process is way longer, althoug we are not sure of active time because we have to come in after hours now to change tapes to complete a backup!

What used to bew accomplished in 1 shift, now takes 24 hrs with extra trips into the office.

Symantec needs to come up with a solution for this as this is unacceptable. If anyone here has found a solution to this problem, let me know!

[HAMZAOUI_Omar]

HI,

Did you try modify block size on drive properties, on "Configuration" Tabs.

[Kevin_Cotreau]

My block sizes are at 64k, which should  be the fastest. In addition, no other changes were made other than the upgrade. The ONLY reason it is slower is going from 11d to 12.5. This seems like they are verifying differently or there is simply a bug.

[ITCowboy]

In our case the configuration is set at 64 KB, which has never changed, and the verify files is turned off. The backups still take too long. The only thing that has changed was the upgrade from 11D to 12.5.

[Kevin_Cotreau]

ITCowboy, I only mentioned the verification because when I compared my 11d logs and my 12.5 logs, the part that took longer was only the verification. Interesting to know that you do not have verification turned on.

[MosesUK]

I also have this issue and ruled out network and hardware issues.  Seem to only achieve around 25% of the performance we were getting before the upgrade.  

Anyone made any progress yet?

[Kevin_Cotreau]

Nothing yet, but if you are getting only 25% of what you got before, that is WAY below what others seem to be experiencing.

[JCasares]

Coming late here but having the same issue as all of you.

We upgraded two 11D installations with different backup hardware to 12.5 SP1 some days ago. The one that took 4 hs to complete until the change now takes 30 minutes more. The other that takes 5 hs, now it's being canceled because it takes more than 7 hs.

Looking at the backup job logs from before and after the change we see every action takes a little more than before and the biggest impact is in the server where the library is attached. How funny! The one that should be supposed to be faster, never slower is taking 50% more time to complete its actions. Of course nothing changed (hardware/software/backed up data) besides enabling AOFO for the task.But I refuse to believe that AOFO can make the task last 50% longer or even more.

What did change with 12.5 to make it slower? We absolutely need to recover some time as we don't want to backup while all people is working.