offline defrag of massive stores

Networking/Security Forums -> Exchange 2000 // 2003 // 2007 & Active Directory

Author: graycatLocation: London, UK PostPosted: Mon Mar 14, 2011 1:38 pm    Post subject: offline defrag of massive stores
    ----
Due to a previously undocumented bug in Symantec's Enterprise Vault we have had our public folders rocket in size to massive proportions. Unfortunately I'm now trying to do an offline defrag of a Exchange 2003 Standard store that is 99Gb.

I've tried the usual eseutil and PerfectDisk neither of which work on the local server even if left for five days or more.

Any suggestions on possible options from here?

Author: AdamVLocation: Leeds, UK PostPosted: Sat Mar 19, 2011 6:42 pm    Post subject:
    ----
5 days!
I guess you must be writing this off to a second file then and not auto-reinstating after it is complete?

First approach will be to see if this is a purely performance problem as to why it is taking so long.

A couple of MS articles explain all the switches and options with ESEUTIL:
http://support.microsoft.com/kb/328804
http://support.microsoft.com/kb/192185
(there's a lot of duplication in there, and I think the first one is probably better worded).
So, using /d /p /t"some file path on a different physical disk" you might be able to improve the speed if you have another set of disks in the system which are on different spindles than the original files. You will still have some contention for throughput on your RAID controller so this may not be sufficient improvement to bring the time down to anything practical, but worth a shot.

Next option: move the files to a newer, better, faster system with better disk throughput, and run eseutil on there instead. This will be especially important if you are running Exchange normally inside a VM, as getting direct access to the hardware with as few layers as possibel is likely to help.
I suspect you might be doing this:
- dismounting the public folders but leaving the mailbox stores running
- running eseutil against the public folders and letting it write the new files back to the same disks
- meanwhile exchange is still using those disks for a small amount of IO
- then in the middle of the night, Exchange does its online defrag of the mail store which increases competition for the disk enormously - you now have two major read and write processes going on

Doing the defrag on another box may avoid this problem, and the time taken to copy the files across may be repaid in better performance.

In all cases you need enough free space for the existing files, plus the new ones, plus 10% head room.

As a rule of thumb, offline defrag seems to run faster if there is more whitespace to be gained. If your database only has a little space to be gained back, it could take a long time to complete.
Make sure you have looked at the deleted items retention period and reduced it if necessary. While you don't describe the bug in detail, it sounds like it has created lots of new items which you have subsequently got rid of, but perhaps they are still being retained.

Hope one, some or all of the above is useful.
Right, back into hibernation now...

Author: graycatLocation: London, UK PostPosted: Tue Mar 22, 2011 12:18 pm    Post subject:
    ----
Holy mother of motherboards ...... AdamV's alive!!?!!

Welcome back, mate. Smile

Thanks for the tips. I looked at the 99Gb store and in the end added a iSCSI target on one of my new SANs and used that. Took about six hours but did the job. Key point to remember: always check you switch / router / firewall settings for any bandwidth throttling between networks first. Embarassed


Unfortunately I've still got a 600Gb PF store on an Exhange 2000 Ent server to consider but I may just migrate the data into EV and then kill it off.

Author: AdamVLocation: Leeds, UK PostPosted: Tue Mar 22, 2011 3:36 pm    Post subject:
    ----
Thought I'd take some time out to give you a chance to catch up with my post count. See you're nearly half way there now... Razz

600GB public folders! On E2000! Eeek! Definitely time for an archive off to EV, I think

Author: graycatLocation: London, UK PostPosted: Tue Mar 22, 2011 5:28 pm    Post subject:
    ----
AdamV wrote:
Thought I'd take some time out to give you a chance to catch up with my post count. See you're nearly half way there now... Razz
Fat chance of me ever catching up, mate! You old timers have just been around tttoooooooo long Wink

AdamV wrote:
600GB public folders! On E2000! Eeek! Definitely time for an archive off to EV, I think


You're telling me!

To make things even better the "undocumented bug" and attempted sneaky fix from Symantec doubled and in some cases tripled every e-mail over 2Mb ....... that wasn't fun! Exchange 2000 then e-mailed these duplicates to the 2007 box which took over a week. Confused



Networking/Security Forums -> Exchange 2000 // 2003 // 2007 & Active Directory


output generated using printer-friendly topic mod, All times are GMT + 2 Hours

Page 1 of 1

Powered by phpBB 2.0.x © 2001 phpBB Group