Should I have separate data and log files for multiple instances?

  • For single instance installations I try to follow a standard configuration where our log files go to E, data files to F, tempdb to H and backups to G.  With a new project we have decided to combine our Test and Dev environments on one server with Test being a default instance of SQL Server and Dev being a named instance.  My plan was to stick with the above convention, but I was asked if it would be better for each instance to have dedicated data, log and tempdb drives.  The problem is we would not have a dedicated controller for each drive.  We would have one controller for the default and named instance data drives, one for the log drives and one for tempdb.  The other issue is having to modify some of my admin scripts/procedures to work with different drive letters.  That's now a real big deal, but does add more work to my plate.

    Do you think it's worth having separate drives for a Test/QA server or is it typically sufficeint to use shared?

    Thanks,  Dave

  • Dave,

    IMHO, todays disk controllers whether fast SCSI, SATA or whatever will be able to keep up with whatever demands you care to throw at them.

    As for the disks attached to them, not knowing the DB size and app I/O requirements once again if the RPM speed is up to modern standards, you "should" be cool to go with your existng drive mappings.

    cheers

  • as a dev, I have to say I'd be pretty unhappy if I had to share my servers with test. 

    I often do stress tests which bog down the server, and QA does the same.  If we had to share we'd be stepping on each other's toes all the time...

    ---------------------------------------
    elsasoft.org

  • Even I agree with Jezemine. It is better to have separate data and log files.

     

    Minaz Amin

    "More Green More Oxygen !! Plant a tree today"

  • To make matters worse they purchased 10k SAS drives instead of the 15k SAS drives I requested.  The server team thought 10k drives would be sufficient because they are SAS drives, however since we planned on sharing the server between development and test, I would have preferred 15k drives.  I'm now trying to find a chart comparing SAS drive I/O rates so I know how much we are losing with the 15k drives.  I know we are losing 50% performance, but how does that translate to I/O rates is what I need to know.

    Thanks,  Dave

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

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