« Previous Next »

Not Answered Thread: application req execution time vs IIS time-taken (for reqs within colo)

Last post 08-03-2009 9:53 AM by Alex80Iis. 3 replies.

Average Rating Rate It (5)

RSS

Page 1 of 1 (4 items)

Sort Posts:

  • 03-31-2009, 1:54 PM

    • redbus
    • Not Ranked
    • Joined on 03-31-2009, 5:27 PM
    • Posts 2

    application req execution time vs IIS time-taken (for reqs within colo)

    Hi! 

    I'm observing a huge time difference(3-4secs) in application's request  execution time and IIS logs time-taken field when there are concurrent requests to IIS. The time difference is huge(2-3secs) even from a machine on the same rack/colo. The application takes 300ms and IIS time-taken is around 3-4secs.

    What metrics should i be tracking to nail down the reason for the delay in IIS and what is the typical solution for this? I have monitored the perfmon's request queue length counter for ASP.NET and its mostly zero.

    Environment: ASP2.0 application, IIS6.0, Win2003 Server edition, Server Grade Hardware.

    Thanks!

    www.redbus.in

     

     

  • 03-31-2009, 7:53 PM In reply to

    Re: application req execution time vs IIS time-taken (for reqs within colo)

    What does the page do? anything else but trigger the app? You could be calling other things and that is why the IIS logs show longer.

    Also make sure you are getting the correct time for the requests.

    In IIS logs you have to calculate the start time. Time-timetaken. As it logs the request at the time when it has finished.

  • 04-01-2009, 1:33 AM In reply to

    • redbus
    • Not Ranked
    • Joined on 03-31-2009, 5:27 PM
    • Posts 2

    Re: application req execution time vs IIS time-taken (for reqs within colo)

    Almost all the requests to the app are served out of cache, there are no external calls. I am using perfmon counters to track the application execution time (request execution time in ASP.NET v2.0----) and also using the System.Diagnostics.StopWatch in our .NET code at the beginning and end of pageload method. The application execution time is quiet small: 300ms. Its the IIS time-taken field that is of concern. Why is it so high when there are concurrent requests (The requests are from a machine in the same colo.) Thanks! www.redbus.in
  • 08-03-2009, 9:53 AM In reply to

    Re: application req execution time vs IIS time-taken (for reqs within colo)

     I experience *exactly* the same problem. Did you ever find a solution? One of my thoughts was that the time-taken is measured until the TCP Ack arrives from the client, as I understand. So would that point us to a network/TCP related problem?

Page 1 of 1 (4 items)
Microsoft Communities