Re: Configuring Disks for Exchange 2003

You might for the database volumes depending on the number of mailboxes in
each and the workload of your users. You shouldn't have any problem with
the log volumes.
--
Ed Crowley
MVP - Exchange
"Protecting the world from PSTs and brick backups!"

"Jason" <Jason@xxxxxxxxxxxxxxxxxxxxxxxxx> wrote in message
news:41BBE288-060A-46ED-8005-0758BF3FBBD1@xxxxxxxxxxxxxxxx
Thank you Mark. Yes we have exchange enterprise. Our current mail
environment is approximately 250GB (using SIS)--most users are well under
50MB but there are enough with 1GB+ mailboxes. We'll also have a
front-end
server to handle SSL etc.

Wouldn't I need to worry about IOPs with only 2 disks per storage group?



"Mark Arnold [MVP]" wrote:

On Mon, 1 May 2006 14:15:01 -0700, Jason
<Jason@xxxxxxxxxxxxxxxxxxxxxxxxx> wrote:

I am planning a migration to exchange for ~500-600 users at an average
mailbox size of 250MB.

After testing with jetstress I found that our existing hardware wouldn't
cut
it (EMC AX100 fiber direct attached SATA SAN) based on the analysis and
horrible 100IOPs.

I've since purchased a direct attached SCSI array with 14-146GB 15K U320
drives and I'm wondering how best to approach dividing the disks.
Here's
what I currently have:

Server:
2-73GB 15K drives in RAID-1 for OS and program
4x146GB 10K drives in RAID1+0 for Logs
Dual channel U320 RAID controller with battery backed cache etc.

Array:
14-146GB 15K drives.


I'm split on whether or not to use both channels and split the backplane
and
have 7 drives on each channel (6 available after global hotspare on each
channel) or leave a single channel and maximize spindles with all 14
drives
(less one for hotspare)?

Thanks in advance for any feedback/advice!

So, you must have Enterprise Edition.
Well, this should be as easy as you want to make it.
The array might look like:
Disk0 - Logs SG1 RAID1 Pair
Disk1 - Logs SG1 RAID1 Pair
Disk2 - Stores SG1 RAID1 Pair
Disk3 - Stores SG1 RAID1 Pair
Disk4 - Logs SG2 RAID1 Pair
Disk5 - Logs SG2 RAID1 Pair
Disk6 - Stores SG2 RAID1 Pair
Disk7 - Stores SG2 RAID1 Pair
Disk8 - Logs SG3 RAID1 Pair
Disk9 - Logs SG3 RAID1 Pair
Disk10 - Stores SG3 RAID1 Pair
Disk11 - Stores SG3 RAID1 Pair
Disk12 - Message Tracking Logs and SMTP Queues
Disk13 - Message Tracking Logs and SMTP Queues

Obviously you get the message and can jig this around to include the
inernal disks.
Use one Store per SG until you max out each of the three stores at 40
to 50GB a throw and then put another store in SG1. If you choose to
use the internals for SG1 and then get as far along the array as SG4,
you will have 200GB of mail before you need to add a 2nd store into an
SG. Clearly you can go do a very large amount of email before you
start worrying about space.
What you will see is the perfect resilience and the sheer number of
disks that could fall over before you actually needed to restore
anyting off tape. You will also see that the IO on the disks is down
to almost nothing (depending on usage, obviously)



.

Loading