none
Need help on explanation of Avg. Disk Queue Length

    Question

  • Based on perfmon, my Avg. Disk Queue Length on physical dick hit 100%.

    What's that mean? Really need explanation

    Wednesday, November 10, 2010 5:31 AM

Answers

  • I’m a bit confused by your statement.   I'm not sure where the 100% is coming from. 

    Avg. Disk Queue Length is the average number of both read and write requests that were queued for the selected disk during the sample interval.

    Current Disk Queue Length is the number of requests outstanding on the disk at the time the performance data is collected. It also includes requests in service at the time of the collection. This is a instantaneous snapshot, not an average over the time interval. Multi-spindle disk devices can have multiple requests that are active at one time, but other concurrent requests are awaiting service. This counter might reflect a transitory high or low queue length, but if there is a sustained load on the disk drive, it is likely that this will be consistently high. Requests experience delays proportional to the length of this queue minus the number of spindles on the disks. For good performance, this difference should average less than two.

    -------------

    This whole topic can get very confusing.

    Think of Current Disk Queue Length as in flight operations.  These are disk read or write that have passes through the Performance Filter Driver and are on their way to the physical disk and back. While in flight a disk activity must pass through (Assuming a SAN) your class drivers, multi path drivers, HBA card the network fabric, Switches and into the SAN.  Any of which could introduce a bottleneck.  Then the acknowledgment of completion must return.

     Think of Avg. Disk Queue Length as disk activities waiting to jump onto the flight.

    So if you have an Ave. Disk Queue Length happening thinks of this as cars backing up on the on ramp to get on to the highway.

    Typically I start disk analysis by looking at:

    Logical Disk\Ave. Disk sec/Read

    Logical Disk\Ave. Disk sec/Write.

    The Queue Length counters are secondary and only used if the latency counters are out of spec.

    Here are some good Blog and tools to use to follow up.

    Taking Your Server's Pulse

    http://technet.microsoft.com/en-us/magazine/2008.08.pulse.aspx?pr=blog

    Performance Analysis of Logs (PAL) Tool

    http://pal.codeplex.com/

    The Case of the Mysterious Black Box

    http://blogs.technet.com/b/clinth/archive/2009/11/18/the-case-of-the-mysterious-black-box-san-analysis-for-beginners.aspx

    Bruce Adamczak


    Bruce Adamczak
    Wednesday, November 10, 2010 2:18 PM
    Moderator

All replies

  • I’m a bit confused by your statement.   I'm not sure where the 100% is coming from. 

    Avg. Disk Queue Length is the average number of both read and write requests that were queued for the selected disk during the sample interval.

    Current Disk Queue Length is the number of requests outstanding on the disk at the time the performance data is collected. It also includes requests in service at the time of the collection. This is a instantaneous snapshot, not an average over the time interval. Multi-spindle disk devices can have multiple requests that are active at one time, but other concurrent requests are awaiting service. This counter might reflect a transitory high or low queue length, but if there is a sustained load on the disk drive, it is likely that this will be consistently high. Requests experience delays proportional to the length of this queue minus the number of spindles on the disks. For good performance, this difference should average less than two.

    -------------

    This whole topic can get very confusing.

    Think of Current Disk Queue Length as in flight operations.  These are disk read or write that have passes through the Performance Filter Driver and are on their way to the physical disk and back. While in flight a disk activity must pass through (Assuming a SAN) your class drivers, multi path drivers, HBA card the network fabric, Switches and into the SAN.  Any of which could introduce a bottleneck.  Then the acknowledgment of completion must return.

     Think of Avg. Disk Queue Length as disk activities waiting to jump onto the flight.

    So if you have an Ave. Disk Queue Length happening thinks of this as cars backing up on the on ramp to get on to the highway.

    Typically I start disk analysis by looking at:

    Logical Disk\Ave. Disk sec/Read

    Logical Disk\Ave. Disk sec/Write.

    The Queue Length counters are secondary and only used if the latency counters are out of spec.

    Here are some good Blog and tools to use to follow up.

    Taking Your Server's Pulse

    http://technet.microsoft.com/en-us/magazine/2008.08.pulse.aspx?pr=blog

    Performance Analysis of Logs (PAL) Tool

    http://pal.codeplex.com/

    The Case of the Mysterious Black Box

    http://blogs.technet.com/b/clinth/archive/2009/11/18/the-case-of-the-mysterious-black-box-san-analysis-for-beginners.aspx

    Bruce Adamczak


    Bruce Adamczak
    Wednesday, November 10, 2010 2:18 PM
    Moderator
  • Just came across this and wanted to add that I think Bonusnet's 100% number came from the graph in Perfmon. It defaults to a max 100 (and clips anything above to that number), and the Avg. Disk Queue Length counter defaults to a scale of 100, so it'll pretty much always look slammed at 100 if you have an ADQL of anything greater than 1. So Bonusnet, the number in Perfmon isn't a % unless the counter has "%" in the name.
    Thursday, February 06, 2014 4:17 PM