Is a normal disk defrag safe for Exchange 2000

[itdamon]

I need to defrag my C:\ and D:\ drives on my Exchange box, but I've read that I should never try to defrag my databases in the D:\Program Files\Exchsrvr\MDBDATA directory. Does anyone know anything about this.

I'm also just planning on using the built in Windows 2000 defrag utility.

Any advice is greatly appreciated.

 

[Zelandakh]

I've been using Exchange for a fair few years. I would never use Windows defrag on an Exchange server...

 

[2cornot2c]

Exchange has its own defrag utility that you must use to defrag the .EDB file. You will need to dismount the exchange store before using the defrag utility. For maintenance reasons it is always good to run the defrag regularly.

Check the following link for instructions on how to defrag the exchange database. Depending on how many mailboxes and the size of your database this process could take hours to complete. Example, we ran the defrag on a 16GB database and it took close to 4 hours to complete. This utility is also very usefull when you have exceeded the size of your database and you need to reduce the database size to re-mount the store.

http://support.microsoft.com/default.aspx?scid=kb;en-us;q192185

 

[itdamon]

I was hoping I was able to word this without getting into offline defrags using esutil or the .edb files. I'm just asking if it is safe to run a normal Windows 2000 disk defrag on a server running Exchange 2000.

I'm versed in Exchange...I'm just wanting to double check that it's OK that I run a disk defrag next weekend on the drive that contains my Exchange database....

 

[2cornot2c]

I don't know if it is safe to do it, but we don't. I also would like to know if it is safe to do it and if it will even do anything for you.

 

[itdamon]

Ok...I'm going to dismount all the mailbox stores...stop all the Exchange services, pull my server off the network and disable my anti-virus. Then I'm going to use the Windows disk defrag utility on my C: and D: drives (no not M:\...I'm not an idiot).

T-7 (time minus seven days)

Somebody let me know in the next 7 days if I'm being a moron...or say a prayer for me, cross your fingers and wait for my update.

 

[xmsre]

If you correctly partitioned the data files for exchange, there would never be a reason to do a disk defrag. Fragmentation happens because of the algorithm used to find space to write to disk. Fragmentation will never occur on a drive that contains a single file. Likewize, fragmentation will not occur when it's all sequential IO, and all the files are the same size. If you keep you logs only on a seperate spindle, all the logs are the same size and IO is sequential; the logs won't get fragmented. The database drive, if it only contains the edb and stm files, should have a very low fragmentation rate. If you seperate the stm and emb files onto seperate drives, they should not become fragmented.

The edb file, internally, stores data in 4K pages. Although pages are freed when items are deleted, the pages are marked free inside the database and the size of the edb does not shrink. Every night, during background cleanup, an online defrag of the database occurs; consolidating free 4K pages. An offline defrage of the edb actually removes the whitespace and shrinks the size of the edb. This is not necessary unless a large amount of mail has been deleted and is never coming back - ie you permanantly removed several hundred users from the server.

 

[itdamon]

Man...that's great info. Unfortunately I don't live in a textbook Microsoft environment. I've been runnning Exchange, IIS, and hosting some asp applications all on the same box and drive.

So the question still remains. Has anyone had any bad experiences using a disk defrag utility on the same drive as an Exchange database?

 

[DG659]

YES was only 50 user but it totally corrupted the .ebd (thank god for veritas and exchange option)

 

[jaksen112]

I've done it, takes a bit longer but I didnt see any corruption (5gb store)

if you think about it, when defrag hits the .edb file, it cant/wouldnt try to open it.. it would treat it like a normal file and move it to a more contig portion of the disk, so how could it corrupt it? maybe if you stopped the process in the middle of the op, but even then I think pointers would take care of that.. just an idea.

01110000