Exchange Server - How do I know how long the defragment will take?

Asked By Lis

03-Mar-08 02:47 AM

Hi,

I use Exchange 2003 on Windows 2003 server. Last night I did defragment to
my email database. The priv1.edb is about 45 GB, the priv1.stm is about 15
GB. The defragment only finished 30% after 5 hours. It seems it would take 10
hours more to complete. I stopped the defragment since my boss want the email
service back up running. I use /p to preserve the old database and use /t to
redirect the temp database to a network drive. My question is how I can know
how long the defragment will take? The other question is what is the .stm
file for?

Your help will be appreciated!

Thanks in advance!

Lisa

03-Mar-08 05:14 AM

It will take 6-10 GB per hour, depending on the speed of your server.
Make sure that you don't click somewhere in the cmd-box during running of
eseutil, or it will stop.
You can start it again with the F5 button though.

The stm-file holds information for non-mapi clients (non-Outlook).
--
Regards,

Rene Frenger
MCITP E2K7
MCP EX5.5, 2000, 2003
MCSE

03-Mar-08 11:13 AM

Network drives are not the fastest thing around; it probably would have had
better performances using a local drive. Besides this, an offline defrag is
anyway quite a slow thing.


Massimo

03-Mar-08 07:59 PM

Hi Rene,

Thank you so much for your reply. Can I pause the defragment process and
resume it later? Let's say if I can stop it and remount the store back, later
dismount it and resume the defragment? It's seems like 10 hours email service
down is not acceptable by my boss.:(

Thanks again,

Lisa

03-Mar-08 08:10 PM

Hi Massimo,

Thanks for your reply. Do you know which is faster? USB hard disk or network
drive? My exchange server doesn't have enough space, so I have to redirect to
another drive.

Thanks again,

Lisa

03-Mar-08 09:21 PM

No.


If you're running the Enterprise verion of Exchange why not create
another mailbox database and move the mailboxes to the new database.
When you're done, dismount the original database and delete the files.

--
Rich Matheisen
MCSE+I, Exchange MVP
MS Exchange FAQ at http://www.swinc.com/resource/exch_faq.htm
Don't send mail to this address mailto:h.pott@getronics.com
Or to these, either: mailto:h.pott@pinkroccade.com mailto:melvin.mcphucknuckle@getronics.com mailto:melvin.mcphucknuckle@pinkroccade.com

04-Mar-08 01:40 AM

It depends on your network and your hardware, but I'd go with a USB (2!)
drive.

Also, as others said, you could create another mailbox store and move
mailboxes there; this would let you move them while the store remains
online, and the new database wouldn't be as fragmented as the old one (but
you'd lose single instance storage, so it could actually grow larger).


Massimo

04-Mar-08 08:55 PM

Why do you think you'd lose SIS? And do you really think it's all that
important if you did?

--
Rich Matheisen
MCSE+I, Exchange MVP
MS Exchange FAQ at http://www.swinc.com/resource/exch_faq.htm
Don't send mail to this address mailto:h.pott@getronics.com
Or to these, either: mailto:h.pott@pinkroccade.com mailto:melvin.mcphucknuckle@getronics.com mailto:melvin.mcphucknuckle@pinkroccade.com

05-Mar-08 02:05 AM

Because Exchange can't store a single copy of a message sent to two
mailboxes if they are on different storage groups? I don't think it's so
intelligent to merge them back together when the mailboxes are moved, one
after another, to the new database.


No, but I warned him he could get an increase in store size; if his users
send or receive a lot of multi-recipient messages, this increase could be
not so trivial.


Massimo

05-Mar-08 09:19 PM

Oh, but it is! SIS is managed by using the message-id value. That
doesn't change when you move a message (it changes if the message is
modified). So moving messages from one database to another will result
in the same SIS ratio if all the messages are moved to the same
database (not storage group).

If you move the messages to a PST and then back to the database the
move back to the database is seen as a new message and a new
message-id is assigned. In this case you do lose SIS.


Back when 9GB disks (and smaller) were "big", this was a lot more
important. Today I'd bet there's more free space in a database than is
saved by SIS.

--
Rich Matheisen
MCSE+I, Exchange MVP
MS Exchange FAQ at http://www.swinc.com/resource/exch_faq.htm
Don't send mail to this address mailto:h.pott@getronics.com
Or to these, either: mailto:h.pott@pinkroccade.com mailto:melvin.mcphucknuckle@getronics.com mailto:melvin.mcphucknuckle@pinkroccade.com

06-Mar-08 03:13 AM

All of this is really beautiful, but I tested it in a lab environment
because I wanted to be sure, and the store size actually increased as
expected.

Steps:

- Create new, empty storage group and database on a test server .
- Put some empty mailboxes in it.
- Have a look at the .edb file size.
- Using one of the users with a mailbox in that storage group, send a
message to all the other ones; use a message with a big attachment (some
MBs), so you can better track what's happening.
- Have a look at the .edb size after sending the message; it will have
increased a lot less than (size of message * number of recipients); SIS is
doing its job here.
- Now create new, empty storage group and mailbox database on the same
server.
- Have a look at the size of the new .edb file.
- Move all the mailboxes to the new database.
- Have a look again at the file size.

I personally tested it on an Exchange 2007 SP1 test server, and it worked as
expected: the new database became much larger than the old one, and its size
was more than (size of message * number of recipients).


Massimo

06-Mar-08 09:07 PM

[ snip ]


Equating the size of the file with the amount of data within the file
isn't always a good comparison. It's also possible that the increase
in size had to do with the inclusion of forms associated with the
messages and not with SIS, or the internal structures in the database
are different. Within the larger database, how much of its size is
free space?

Even if SIS is lost, the effect on a production machine is minimal.
We've move tens of thousands of mailboxes on hundreds of servers and
not experienced what you cite. But we've never done it with such a
small amount of data and such a limited number of mailboxes.



--
Rich Matheisen
MCSE+I, Exchange MVP
MS Exchange FAQ at http://www.swinc.com/resource/exch_faq.htm
Don't send mail to this address mailto:h.pott@getronics.com
Or to these, either: mailto:h.pott@pinkroccade.com mailto:melvin.mcphucknuckle@getronics.com mailto:melvin.mcphucknuckle@pinkroccade.com

07-Mar-08 03:05 AM

Correct. That's why I used test mailboxes containing only that message and
test stores containing only those mailboxes ;-)


I agree with this, but I think that, depending on the usage pattern of the
Exchange server, moving all mailboxes to a new store actually *can* cause a
size increase. Think about a server which has a lot of distribution lists.


Massimo