Sharepoint 2010 and how to Baseline using sqlio and sqliosim

  • Good afternoon!

    OK, We are rolling out sharepoint 2010 on sql server 2008 in the next couple of months and the microsoft tech we have on site is asking for a baseline from either sqlio or sqliosim for our environment. I am trying to figure out how to do this and present the data properly, I have searched the net, and ran through Brent Ozars docs, kevin klines blogs .. but they don't seem to match my environment and I am anything but overly knowledgable as far as measuring and reporting on IO. We have a san and we are connecting to it through duel fiber cards. We are on a dual node cluster and our storage guys have given me storage on an"F" drive and then individual partitions under it like this

    F:\TempD1 (5gig)

    F:\TempD2 (5gig)

    F:\TempL1 (5Gig for log file)

    F:\SearchD1 (40gig)

    ...

    ..

    and so on we are giving each content database a data folder and a log folder each on their own drive. We are giving the Temp Databases 4 separate folders (one per cpu thread) and 1 for the log, the search database 4 for data and one for logs, so all said and done I have 20 or so separate drives just for sharepoint. Now when doing SQLIO should I be testing one drive at a time.. or should I just make one giant parameter file large files on each drive...

    Also if this setup is totally ridiculous and overkill I won't mind hearing that as well, This is our first delve into the sharepoint world and are always interested in others successful (and unsuccessful) implementations as well as best practices that we may have not seen.

  • I can't speak of Sharepoint 2010 yet, but 2007 can hammer a sql server.

    Dividing into files per core isn't recommended. I know that it's a common practice for temdb, but it's a "blind" optimization. Some guy did 30 sql myths in 30 days and he had a script to tell whether you needed more files than you already had. A better approach. Also the sizing really depends on how sharepoint are going to be used. Placing all databases on the same drive can result in a performance bottleneck depending on how it's done whether using mountpoints or not. Also have a chat with the storage guys about whether you need high performance, low performance, sequential or random io.

    Testing with sqlio with one file at the time will only tell you the performance of one file, make the test as close to your expected production environment as possible.

    Our website is based on Sharepoint 2007 and see a steady stream of 20mbit data.

  • Testing with sqlio with one file at the time will only tell you the performance of one file, make the test as close to your expected production environment as possible.

    😀

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

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