Planning Disk and File layout

  • Hi all

    We are looking to migrate from a busy datawarehouse server to a new server and I have been tasked with ensuring the database files are all on appropriate disks that are yet to be determined.

    I am thinking about capturing IO using sys.dm_io_virtual_file_stats to work out those files most heavily used, and also bench marking the new disks with SQLIO and working out from there the appropriate file to disk lay out.

    What do people think of my approach, does anyone have any tips for doing this?

    Many thanks

    P.S. I know we should probably be bench marking our servers in the first place and this is the start of that exercise.

  • That's probably the route I'd go down. Would also be interested in other approaches if anyone has any.

    Thanks,

    Simon



    MCSE: Data Platform
    MCSE: Business Intelligence
    Follow me on Twitter: @WazzTheBadger
    LinkedIn Profile: Simon Osborne

  • First question to ask yourself during data warehousing (talking from my own experience here)...

    How strong is your server? Then, look at your disk speeds and set-up of your disk array.

    remember a fully populated array works faster than a half full array. Also separate your dimension loads and fact loads.

    I personally had an increase of around 60% odd but simply playing with these variables 🙂

    Looking at loading stats for me did not help, cause we all know...every server has its own demons 🙂

Viewing 3 posts - 1 through 2 (of 2 total)

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