2

I have a Exchange 2007 SP1 server running on Windows 2008

6 HD Drives in a RAID-1 OS, DB, Logs on separate RAID-1 Disks

Size of the Mailbox Database is 183GB and increasing

We only have First Storage Group and Second Storage Group

There is no more space on the server to install new Physical Disks and create a Storage Group

Q - Can I resize the RAID-1 Partition where the DB is ?

Q - Any other suggestions as to how I can decrease the Mailbox DB Size ?

Will be grateful for your suggestions on this.

Kind Regards

sysadmin1138
  • 133,124
  • 18
  • 176
  • 300
Mutahir
  • 2,357
  • 2
  • 32
  • 42
  • 1
    For most businesses, e-mail is a critical system; and yet it seems common for management to treat it like an optional system. Critical systems cost, and keeping them running smoothly costs again. – Chris S May 21 '10 at 22:19
  • Chris S, Yes - we are very well aware of that - this is not the question. I answered with a Plan on upgrading the Disks (not talking about costs) - rather talking about Exchange 2007 over here. We can even invest in a new server if we have to - so cost isn't an issue - I don't know where in my question did you got an idea that we have cost issues :-) More sort of a technical question – Mutahir Jun 02 '10 at 15:35

3 Answers3

3

Good news is that Exchange 2007 SP1 has database portability, meaning you can dismount the databases, move them and remount them with no ill effects other than down time. ( http://support.microsoft.com/?id=555603 )

Bad news is that your company will have to spend money.

The biggest challenge for Exchange with large datastores is I/O. You can get huge drives now (like 2TB) but the random I/O of Exchange will quickly swamp most storage. So you need to split your datastore into multiple smaller DS spread out across more disks. Expensive.

Best solutions is to get another server and dedicate it as a mailbox server only and keep you existing server as a CAS server. Again, expensive.

I think I would look at getting an add-in SCSI controller and adding a set of external hard drives.

Another cheaper option (although slower) is to move the datastores off to iSCSI storage made from a box running some free iSCSI software like FreeNAS.

Whatever you decide, do it before you run out of room!!

HTH

Mike

BhmJeep
  • 156
  • 2
2

If you are out of space, it's probably time to run in circles scream and shout. If you have space but its on the "wrong partition" you could set up another information store on a new storage group on the other partition and move some mailboxes there. After you have some breathing room it's time to sit down with management and decide whether to get bigger disks to resize the partition (if that's possible with your raid controller) get another server, or decided to create and enforce some retention policies and quotas.

Jim B
  • 24,081
  • 4
  • 36
  • 60
  • There is no more physical space on the server to install new Physical Disks and create a Storage Group I still have 92GB Left on the DB Hard Disks (Raid-1-partition), other partitions are fine like OS and Logs. I am worried as the size of the DB is increasing, what would be the best approach to be prepared. Is it possible that I buy two 750GB Disks, then take ONE Live disk out of the DB Partition, insert one of the 750GB, re-build the mirror and then take the 1st low space disk out and insert the 2nd 750GB ? will that increase the space ? Thanks for your suggestions. – Mutahir May 18 '10 at 08:05
  • 1
    Messing with the RAID trying to swap in 750GB disks is asking for trouble. It's a better idea to use imaging software to make an image, then swap both disks, and write the image back. That way you have two ready-to-go disks if something goes wrong. Messing with the RAID could hose your disks, it's unlikely, but why take the chance? – Chris S May 21 '10 at 22:18
  • 1
    Making an image isn't a bad idea but has it's own drawbacks You would have to shut down exchange while you make an image and if the software doesn't work with the raid controller you have you take a hit for even more downtime. Also if you don't think the raid controller will rebuild the mirror why would you bother trusting it with your data in the first place – Jim B May 22 '10 at 02:11
  • Backup the databases. swap one drive, rebuild the mirror, swap the other rebuild again, expand the partition from Disk Manager. that IS the correct way to do it. For the raid 5, it really depends on how many drives, as rebuilding it, say 12 times for 12 drives is way to hard on the HDDs and needs to be made in stages. A 4 drive synology/qnap could be used to TEMPORARILY move the databases to a SMB share on it. swap the raid 5 drives and copy them again. you could use iscsi but it is not worth the effort for a temp solution – Rostol Jan 26 '19 at 06:50
0

thanks for your reply, i have created a plan as follows :

1 - BESR 2010 - Image the whole server, test baremetal restore on another box and also test virtual conversion to hyper-v (just to be sure).

2 - Shutdown the exchange box (Arranging a downtime), replace the hard disks for DB partition only (2 disks) with larger capacity

3 - Create raid on those two disks (RAID-1)

4 - As I won't be replacing the OS Disks, I will just use BESR to restore the DB to the new raid-1 set (new disks - larger capacity).

5 - Once the restore is completed, reboot the box

Do you think that would work ? BESR 2010 flushes the logs but we have circular loggin enabled - it does howerver stamps the exchagne db as backed up and in Clean state.

What do you recommend.

Mutahir
  • 2,357
  • 2
  • 32
  • 42