locked
Disk Transfer (reads and writes) Latency is Too High - Exchange 2007 Servers RRS feed

  • Question

  • Hi All,

     

    We've been getting a lot of alerts indicating "Disk transfer (reads and writes) latency is too high" for our new Exchange 2007 server. The alerts are being generated on our active primary node of a CCR cluster. The alerts used to be generated primarily in the evening hours during backups, but we're finding that as we migrate more users to the Exchange server, the alerts are becoming more frequent and even happening during the day. The alerts are generated primarily on the SAN storage used to store the mailbox databases and log files.

     

    Here are some sample alerts I'm getting:

     

    The threshold for the Logical Disk\Avg. Disk sec/Transfer performance counter has been exceeded. The value that exceeded the threshold is: 7008.28408655617.

     

    The threshold for the Logical Disk\Avg. Disk sec/Transfer performance counter has been exceeded. The value that exceeded the threshold is: 2299.8383140591.

     

    The threshold for the Logical Disk\Avg. Disk sec/Transfer performance counter has been exceeded. The value that exceeded the threshold is: 118870.54512909.

     

     

     

    Am I correct in understanding that the value is the time in seconds and that if the value exceeds 50 seconds over 5 samples, then an alert is generated? What performance graphs should I be viewing in SCOM that helps identify if there is a problem with our SAN backend storage versus a threshold override that I should be putting in place?

     

    I want to make sure that if I have the threshold value set to low, then I want to make sure that I can see what our average is and then set the threshold accordingly, but if there are graphs that comes with SCOM that I can view that indicates there is a performance problem with our disks, then I want to make sure that gets addressed accordingly.

     

    Many thanks in advance!

     

    Edward

    Wednesday, April 14, 2010 12:47 AM

Answers

  • Hi,

     

    Based on my research, I would like to suggest the following:

     

    1.   Check if there are disk performance issues on the system with performance monitor.

     

    2.   Adjust the Thresholds settings of the monitor.

     

    Please also refer to:

     

    Overview of Available Monitor Types

    http://technet.microsoft.com/en-us/library/dd440880.aspx

     

    Hope this helps. Thanks.


    Nicholas Li - MSFT
    • Marked as answer by Nicholas Li Monday, May 3, 2010 2:47 AM
    Thursday, April 15, 2010 7:23 AM
  • This monitoring is done via the OS Management Pack. It looks at the counter for 5 samples, if it's over the threshold for 5 samples (minutes) it will alert.

    I could not find this counter in a view anywhere. What you can do is to go to the "Logical Disk State" view, right-click on the disk you're interested in, select "open" then "performance view". It will list all counters collected for the disk. Then select the correct one.


    This posting is provided "AS IS" with no warranties, and confers no rights. Use of included script samples are subject to the terms specified at http://www.microsoft.com/info/cpyright.htm
    • Marked as answer by Nicholas Li Monday, May 3, 2010 2:47 AM
    Friday, April 16, 2010 1:37 AM

All replies

  • Hi,

     

    Based on my research, I would like to suggest the following:

     

    1.   Check if there are disk performance issues on the system with performance monitor.

     

    2.   Adjust the Thresholds settings of the monitor.

     

    Please also refer to:

     

    Overview of Available Monitor Types

    http://technet.microsoft.com/en-us/library/dd440880.aspx

     

    Hope this helps. Thanks.


    Nicholas Li - MSFT
    Thursday, April 15, 2010 7:23 AM
  • Hi,

     

    Based on my research, I would like to suggest the following:

     

    1.   Check if there are disk performance issues on the system with performance monitor.

     

    2.   Adjust the Thresholds settings of the monitor.

     

    Please also refer to:

     

    Overview of Available Monitor Types

    http://technet.microsoft.com/en-us/library/dd440880.aspx

     

    Hope this helps. Thanks.


    Nicholas Li - MSFT
    • Marked as answer by Nicholas Li Monday, May 3, 2010 2:47 AM
    Thursday, April 15, 2010 7:23 AM
  • This monitoring is done via the OS Management Pack. It looks at the counter for 5 samples, if it's over the threshold for 5 samples (minutes) it will alert.

    I could not find this counter in a view anywhere. What you can do is to go to the "Logical Disk State" view, right-click on the disk you're interested in, select "open" then "performance view". It will list all counters collected for the disk. Then select the correct one.


    This posting is provided "AS IS" with no warranties, and confers no rights. Use of included script samples are subject to the terms specified at http://www.microsoft.com/info/cpyright.htm
    • Marked as answer by Nicholas Li Monday, May 3, 2010 2:47 AM
    Friday, April 16, 2010 1:37 AM