BackupExec and my servers

[MK-2]

We have one tape loader, connected to the admin server which has backupexec. the academic backup job, which is 140GB, runs in a little under 6 hours and completes without any probs (well, that was after i deselected the windows folder to be backed up, previous guy just selected everything!).

now the admin backup is only about 65GB, but so far its been running for 10 hours and has done about 45GB. I've looked at the job monitor and it's going dead slow. Obviously taking this long isn't good as it means the academic backup doesn't run until about 10hours after its schedule time, making it run during the day.

Exchange is on the admin server and I've noticed that it sometimes slows down when it hits the mailboxes, but surely that wouldn't take it down to this speed.

Could it be if people are using outlook web access that backupexec cant write a file or something, if its open by someone else?

 

i just want to sort this backup out as it was about may the last time any backups ran successfully!

[nicholab]

Do you have live update enable on backup exec an update could have knackered it/or make sure all update have applied some need restarts? Can you do an incremental backup? You are running the exchange agent not just the standard also have you thought of upgrading to 12d?

[MK-2]

i'll have a look at live update, never thought of checking that.

these jobs were set in place before i started, so even though they are full backups, i dont want to start messing just yet until i can confirm with people exactly what needs backing up (as the whole c: and d: are being backed up at the mo!).

we will be getting 12d in october, but until then its got to be a case of make do.

 

i'll look next week at creating new backup jobs that do only essential data, but the problem is i'll still have to back up mailboxes and thats where its slowing to a crawl.

i was wondering perhaps if it could be bad sectors on a drive or something, that its taking an age to read the data?

[matt40k]

I normally have exchange on it's own server, like sims as it's a memory hog. I've always used Backup Exec (latest) with exchange pack. I would assume it's a memory issue, what version of Windows are you running and what about of RAM do you have?

[MK-2]

its on 2003 r2, and has 4gb.

[matt40k]

Standard or Ent?

 

I think Standard is limited to about 2gb of ram per app or less. I could be wrong.

 

It might be worth getting another 4gb ram and going enterprise. However I wouldn't without checking out with someone else.

 

I would prob setup performance monitoring to see if you can pickup the bottleneck.

[SYNACK]

It may not directly solve the throughput problem but may help to minimize it:

 

Have you run a defragment on the private mail store using eseutil it should order and restructure the mailbox database on the server and make it a whole lot quicker to copy and hopefully to backup as it won't involve a whole lot of extra seeks that would add to the time taken to back it up.

 

You can use the same eseutil program to check the integrity of your DB, I would do this before the defrag. The DBs can get corrupted a small amount only effecting a few user mailboxes, the DB still mounts but it may be causing some of the issues with the delay if it is repeatedly trying to read correct data from any corrupted accounts inside the DB.

 

Full command help here:

Eseutil

[MK-2]

i'm looking at redoing all the backup jobs (creating the jobs).

i just noticed that the jobs are all full backups not incremental. especially in terms of mailboxes, it seems a waste to continually backup full mailboxes when the data wouldn't have changed. i'll have a look at the db tool too, as i'm sure someone said they archived some emails and on the next backup job it listed them as corrupt files so it could be that too.

i'm also going to put both academic and admin jobs on to one tape. i mean we have the AIT5 tapes and at most use 200GB across both backups.

 

cheers for the help guys