Another Raid config question

  • Hello

    I currently have the following disk config on a server that is used to run a financial/erp system

    2 x 74Gb disks - Raid 1 - Main DB logs

    4 x 74Gbb disks - Raid 10 - Main DB Data

    4x 74Gb disks - raid 5 - Backups

    2 x 74Gb disks - Raid 1 - Temp DB

    This database is about 70gb in size and have about 10gb of blobs with lots more blobs to come.

    We are installing a new server and have bought 14 x 146gb disks for the DAS and I was wondering if I should stick to the current design or tweak it. Looking at the current setup I cannot see why they went for a 4 disk raid 10 config (2 spindles) .

    What I am thinking at the moment is

    2x 146Gb - Raid 1 - Main DB logs

    2x 146Gb - Raid 1 - Tempdb

    4 x 146gb - Raid 5 - DB Backups

    6 x 146GB - Raid 10 - Main DB Data

    Any suggestions ?

    TIA

    Simon

  • conceptually looks good but I'd be afraid of running out of space.

    You stated that your DB is ~70GB today. How fast is it growing ?

    what kind of activity is on the DB ?

    If you use TempDB heavily or if you log heavily (During index defrags for example), you could see some issues here.

    if you had more small disks you'd be better off using RAID 1+0 across the board, but generally that's not an option due to price.

    this may be as good as you can do for now.....

    monitor IO and disk space and hopefully you'll be fine.

    Gregory A Jackson MBA, CSM

  • Also - Backups are just about the heaviest load you can put down on storage volumes. Since RAID-5 is penalized for writes - I wouldn't use that on the backup drive (as a matter of fact - I'd put logs or the DB files there first before I use RAID-5 on the backups drive).

    Of course if it's not being hammered continuously - it might work for a little while.

    ----------------------------------------------------------------------------------
    Your lack of planning does not constitute an emergency on my part...unless you're my manager...or a director and above...or a really loud-spoken end-user..All right - what was my emergency again?

  • do you have any stats and analysis for how your databases are used ? If not then you're just making pure guesses. So many want to split tempdb but put it on less spindles than the main datafiles thereby adding a bottleneck if tempdb is used. I think you'd probably do better by making your main raid into all the disks for tempdb and data and putting tempdb on multiple files ( 1 per core )

    Raid 5 has no use except for read only filegroups - need to post so i read your original post again!

    [font="Comic Sans MS"]The GrumpyOldDBA[/font]
    www.grumpyolddba.co.uk
    http://sqlblogcasts.com/blogs/grumpyolddba/

  • yeah a 4 disk raid 5 gives write performance = to 1 single disk, you'd get roughly 4 times write performance making it a 4 disk raid 10.

    [font="Comic Sans MS"]The GrumpyOldDBA[/font]
    www.grumpyolddba.co.uk
    http://sqlblogcasts.com/blogs/grumpyolddba/

Viewing 5 posts - 1 through 4 (of 4 total)

You must be logged in to reply to this topic. Login to reply