Calculating Avg.Disk Queue length when the storage is on SAN

  • Hi,

    We are using perfmon to monitor the sql instance. The storage for this sql instance is from SAN.

    we have the disks as below:

    C- OS

    D- data files

    E- log files

    T- TempDB

    Z- backups

    Each disk has 150 spindles in it.

    Perfmon giving High Avg.Disk Queue length alarms of values up to 33 some times.

    So is Perfmon takes consideration of the no.of spindles in each disk & then measures the Avg.Queue length? or

    do we need to divide Avg.Disk Queue length that Perfmon giving by the no.of spindles in each disk? for example 33/150 = 0.22 , which <2 & has no disk contention right?

    Please clarify me

  • rambilla4 (4/6/2010)


    Hi,

    Perfmon giving High Avg.Disk Queue length alarms of values up to 33 some times.

    So is Perfmon takes consideration of the no.of spindles in each disk & then measures the Avg.Queue length? or

    Please clarify me

    Perfmon has no idea, the OS simply see's each disk as an individual volume, it does not know what that volume consists of. Each disk is simply a disk to the operating system.

    /* ----------------------------- */
    Tochter aus Elysium, Wir betreten feuertrunken, Himmlische, dein Heiligtum!

  • Perfmon has no idea, the OS simply see's each disk as an individual volume, it does not know what that volume consists of. Each disk is simply a disk to the operating system

    So is the below formula correct?

    actual Avg.Disk Queue length = Avg.Disk Queue length value from perfmon/no.of spindles in disk

    thanks

  • The question about perform's avg disk queue length and its usefulness has come up before on this forum. Here is one of the other conversations about it with a couple of good tips:

    http://qa.sqlservercentral.com/Forums/Topic823698-146-1.aspx

    There are probably others as well. If I were you I'd start by searching for other ways to measure disk subsystem performance in regards to SQL Server besides avg disk queue length.

    Steve

    SQL Managed

  • ADQL is completely useless as a disk performance metric in the modern age. Use avg disk sec/read and avg disk sec/write instead. Even better, use the SQL Server DMV sys.dm_io_virtual_file_stats to determine the IO stalls from SQL Server's perspective - that is what truly matters anyway.

    Best,
    Kevin G. Boles
    SQL Server Consultant
    SQL MVP 2007-2012
    TheSQLGuru on googles mail service

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

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