RAID setup recommendations for SQL Server machine.

Posted on 2007-03-28
Last Modified: 2008-03-03
I am looking for recommendations on what type of RAID system we should implement for this server and (why).

Here is the situation:
1. Windows NT 2000 server - Applications: SQL Server 2005, Great Plains, miscellaneous utilities (our Exchange / Mail is a different server)
2. All SQL server Tables, GP tables and data
3. All user's (50) document folders, they store all their documents on this server.

Here's what I would like to achieve:
1. If we crash, instant or easy restoration of the operating system, all configuration, SQL Server and GreatPlains configurations.
2. Easy restoration of 'data' files.

We already own:
Adaptec 3200s, 2 channel Scsi RAID controller
8 identical 32gig SCSI drives .

My 3rd party network vendors are suggesting 2 Arrays of 4 drives each (32gigs), both would be RAID LEVEL 5.

Channel 1would have the OS in one partition, SQL server and GP in another.
Channel 2 would have 'data' files only in one partition, a second partition for 'other' stuff.

Can / should I 'ghost' or image the operating system and SQL installtion to another drive on our network as a precaution?  Configuration does not change from day to day.

Question by:Volibrawl
  • 3
  • 3

Accepted Solution

dglenp earned 400 total points
ID: 18810036
I set up both my SQL Server and Exchange Server boxes with RAID 1 and RAID 5.  In your scenario, I would set up a RAID 1 array using 2 drives for the OS and program files (I also put the SQL transaction logs on this array).  Set up a RAID 5 array using 4 drives for data (SQL/GP data and user files).  Configure 1 of the remaining drives as a hot spare that can be immediately put to use by the controller in the event one of the other drives fails.  Put the 8th drive on the shelf as a spare part.  I recommend this mainly because of the spare parts issue.  I had a RAID 5 array with three drives go down once when TWO of the drives failed within seconds or minutes of each other.  The 4-drive array for data is a better setup.  But I think a 4-drive RAID 5 for the system partition is a bit over done.  I'd prefer to have a spare drive on the shelf.  IMHO.

The 'ghosting' issue depends on your down-time tolerance.  With RAID arrays and spare drives, your data is fairly secure...unless the 3200s controller or the motherboard fails.  Then all the happy data on your fault-tolerant RAID arrays is useless unless you have spare parts at hand or a stand-by server available.  You say you want to achieve "instant or easy restoration."  You will have to consider whether your environment requires high availability options such as server clusters.  As far as recovery goes, if you have spare parts and good backups, you can be back on line having lost little or no data.  However, in a worst case scenario, without high availability, it may take hours to do (or a day or two if you have to order a new server or motherboard, etc.).


Author Comment

ID: 18811185
Thanks  ...

Your suggestion is well-taken and where I would like to go with my setup.

Right now, my network guys have returned the server (bootable), with updates, Backup tape, etc.
The setup is 2 RAID 5's.  OS is on the first Array
Data (preserved) is on the second Array.

Is there a way to "break up" Array0 (4 drive RAID5) so I can re-configure it as 2 RAID1's ? (consistent with your idea adn M$ suggested setup for GP.)

Bear in mind:
1. I don't want to have to re-install NT and all the updates from scratch
2. I don't want to chance losing my data (on Channel2 RAID5).

M$ suggests:
RAID 1 for operating system and applications (2 disks)
RAID 1 for SQL database log files (2 disks)
RAID 5 (4 disk minimum) for SQL data files
RAID 1 for TempDB (2 disks) - OPTIONAL, but recommended - maybe later ...

Expert Comment

ID: 18811595
If you split the 4-disk RAID 5 into two RAID 1's, you will lose the data.  You may already know that RAID 5 requires at least 3 drives to set up.  It can run with degraded performance if one drive fails, but it takes 3 to set up.  RAID 1 is simply mirroring, so it only requires 2 drives.  To re-configure the server you could Ghost the channel 0 volume or make a complete backup using some other tool, re-configure the array, then restore the image onto one of the RAID 1 arrays and make it the boot volume.  The backup and restore would be good disaster recovery practice anyway.

Through all of this, the data on the other volume wouldn't be disprupted.
Optimizing Cloud Backup for Low Bandwidth

With cloud storage prices going down a growing number of SMBs start to use it for backup storage. Unfortunately, business data volume rarely fits the average Internet speed. This article provides an overview of main Internet speed challenges and reveals backup best practices.


Author Comment

ID: 18812310
Thanks, that is exactly what I am doing now...

1. Mirroring our current C: drive onto an IDE drive.
2. Mirroring our current 2nd RAID array (Array1 RAID5- Our data) onto the same IDE drive (for failsafe BU).

Once we are comfortable that we still have all our data on the IDE, we can break apart our OS RAID5 and reconfigure it into the 2 RAID1s.

Then we will try to boot from the IDE and mirror the OS back to the newly created  Array.
Working on it right now  ..

I need to look at "how to ghost and restore from Ghost".  Will be my next question...

Expert Comment

ID: 18813519
Depending on which edition of Ghost you have, it should allow you to boot from the CD and copy from volume to volume.  You may have completed the task by now, though.  It's late here.  Just checking the update...

Assisted Solution

_Mr_Limo earned 100 total points
ID: 18821540
Best way to minimize possible down time is to have the 2 RAID 5's - one with the OS (3 drives) and the other (4) drives as the data - a RAID 1can and will most likely fail - if the failure of the one drive causes Media Damage on the other drive at the same time, you have 2 dead drives (it happens all the time).  in a RAID 5, which provides the most fault tolerance, you can have 1 (ONE) drive fail in the array & not lose the use-ability of the device.  It's not possible for 2 drives in a raid 5 to fail and NOT lose data.  Always have an alternate backup - a current Ghost image is always your best bet.  RAID 1 does not give enough fault tolerance for the o.s., if it's a necessary server.  

Author Comment

ID: 18823278
Always anxious to hear logically supported recommendations .....

Featured Post

Portable, direct connect server access

The ATEN CV211 connects a laptop directly to any server allowing you instant access to perform data maintenance and local operations, for quick troubleshooting, updating, service and repair.

Question has a verified solution.

If you are experiencing a similar issue, please ask a related question

Patch Management is administrated by page 129 of  the following document: However, there are issues with Patch Management with FTP (for instance, a slow Internet connectio…
this article is a guided solution for most of the common server issues in server hardware tasks we are facing in our routine job works. the topics in the following article covered are, 1) dell hardware raidlevel (Perc) 2) adding HDD 3) how t…
Finds all prime numbers in a range requested and places them in a public primes() array. I've demostrated a template size of 30 (2 * 3 * 5) but larger templates can be built such 210  (2 * 3 * 5 * 7) or 2310  (2 * 3 * 5 * 7 * 11). The larger templa…

839 members asked questions and received personalized solutions in the past 7 days.

Join the community of 500,000 technology professionals and ask your questions.

Join & Ask a Question