Video Screencast Help
Symantec to Separate Into Two Focused, Industry-Leading Technology Companies. Learn more.

Backup exec 2010 R3, Slow differential Backups

Created: 17 Jan 2013 | 6 comments

Hi,

   A little stage setting first:-

we have Backup exec 2010 R3, running on our backup server (MS Windows server 2008 32bit, 2 intel 3.06Ghz Xeon processors, 2G ram), Full backups transfer rate varies between typically 450-1,800MB/min, to a ESATA connected WD My Book Essential (wdbacw0030hbk-01 3TByte drive).

Unfortunately due to the length of time backups take(full backup off all our servers takes all weekend!), we've elected to do differential backups every weekday overnight.

The BIG problem is the differential backup of our main fileserver is very slow 20-50MB/min. I've had a look at the backup logs and one obvious issue seems to be the very large numbers of folders its backing up, for example:-

 

Byte count             : 12,697,968,916
bytesRate              : 31.00 MB/Min
Files                     : 116,528
Directories            : 1,326,746
Skipped files          : 0
Corrupt files          : 0
Files in use           : 0
 
Q.How come the directory count is way higher than the file count?
 
Any insight into the issue would be great,
Thank you!

Comments 6 CommentsJump to latest comment

pkh's picture

The job has to process each file and directory even if they are not backed up.  Even if there is nothing to backup in a directory, the structure is backed up.

If you are using the archive bit method, try using the modified time method with the USN journal.  This should speed things up.  When you switch method, make sure that you do a full backup before doing your differential backups.

MrRubbs's picture

Ha!

That makes sense, the number of folders doesn't change by much, wherehas the files do,folder size is small and without some better defragger, the folders will be all over the place! Hence lots of slow disk access!

How does the Modified time method improve the situation, and will changing it invalidate previous backups done with the archive bit method?

Thanks of the prompt reply! 

pkh's picture

With modified time, especially with the USN journal, BE only have to access the USN journal to determine which files to back up, whereas for the archive bit method, each file needs to be examine to determine whether the archive bit is on or off.  Hence the modified time method should be faster.

The previous backups are not affected when you switch method, but you got to do a full backup after you switch methods.  You can just continue with your incr/diff backups.

MrRubbs's picture

Hi,

   Sorry for the delay, this weekend was first chance to try the alternative backup method, took slightly longer for the full backup(50 hrs rather than 48).

Big problem is the differential backup, it would normally complete overnight, but it's nowhere near finishing and I've had to cancel it, it almost looks as if it's doing a full backup again.

I've tried investigating the issue(searching for similar issues), and one advisor mentioned that a recreation of the file selection lists made need doing.

Q. Why is this necessary?

Q. Is there an easy way of using the existing Text based selection lists to refresh/update the lists?

Doing it by hand for all the lists, with the number of server we have would take quite some time!

Thanks again for any help,

MrRubbs's picture

Hi,

   I've recently discovered that our AV (Sophos Antivirus), may be causing issues with the backups.

We don't run it in "on-access" mode on the servers, BUT the overnight scheduled AV scan is very likely changing the USN journal (search for 43898 (USN Change Journal And Sophos)), thus ensuring our daily differentials would include everything it scanned!

       I've followed the article mentioned to disable it's altering of the USN, but will not know for sure until we've had a "undisturbed ", full back, followed by a differential.

Hope this makes sense!

Dave.

MrRubbs's picture

Hi,

I Hope someones still able to help!

The full backups are completing, BUT the differentials are still not completing overnight!

How on earth do I work out if the USN Journals are still being altered, or is there some other factor that needs to be factored in.

At this rate we'll probably ditch Backup Exec, and pay for someone to come in and quote us for a deduplication backup solution that actually works!

This is so frustrating, and from reports about BE2012, I won't be considering upgrating, come back Peter Norton save us from this lingering death......

(Sorry about this BUT this is why technology just gets a bad reputation...it's too damn complicated!)

Rant over!  Anyone kind enough to offer some guideance??  Please.

Dave.