• Experiance has shown ( for me ) that where the sql servers are connected to a company SAN used for many other servers the performance is seriously lacking compared to DAS. Contention from shared disks has proved to be a problem although it took much digging before the physical setup of the disks was admitted.

    Performance counters are very difficult as there tends to be a reliance upon the SAN provider to provide this data and they appear to be reluctant to ever accept there may be an issue.

    I have monitored trends but the counter values themsleves are valueless when compared to their real values with DAS, this makes it difficult for example to calculate the i/o limits - I have a 5 spindle raid 5 ( shared ) I can record values into the thousands for the i/o and into the 100's for the queues - I know these values are meaningless and I should only take them as an indication but what would be a baseline and what does a disk queue value of 268 actually represent?

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