Time increases for processing

Home Page › Forums › BizTalk 2004 – BizTalk 2010 › Time increases for processing

Viewing 1 reply thread
  • Author
    Posts
    • #18294

      Hello, 

      I am having an issue with BizTalk taking longer each time it processes a transaction, and I am hoping some of you with more knowledge of the interworkings of BT and SQL can help me identify where the issue is. 

       This is a simple HTTP receive, with 2 subscriptions.  One is a file drop, which just passes the doc to the file system for backup, and the other is an Orchestration.  The Orchestration passes the file to a WS, gets a response, and sends the sync response back to the TradingPartner.

      Architecture:

      BizTalk 2006 installed on one server out in the DMZ, and SQL installed inside the cort domain.  This is the first time I have installed BT and SQL on seperate machines. 

      Symptoms: 

      1. In my IIS log, the time taken increases about 3 seconds per transaction throughout the day.

      2. Usually in the morning it resets to 1-5 seconds, and periodically throughout the day, but I have not identified what resets it.  I have tried IISreset, restarting the BT host, and restarting SQL. 

       3.  In HAT, the time lag seems to be from the first page to when the subscriptions actually kick off.  The passthru subscription only takes a second, and the Orch Subscription only takes 1-2 seconds. 

       

      This leads me to think that there is some sort of logging or caching that is going on between the BT host and SQL if I understand the times in HAT.  Maybe this logging or caching resets itself periodically causing the times to drop at night.  Other than that I cannot pinpoint anything else going on at the times it drops back to 3 seconds using the App logs. 

       

      Any suggestions or tools as to where to start would be greatly appreciated.

      Thanks

      mike

       

    • #18310

      Finally I figured out why the time taken periodically drops back down to 3 seconds.  The App Pool in IIS 6 is set to Shutdown Worker process after 20 minutes, so when the partner does not send anything for over 20 minutes, it resets.  To verify, if I recycle the App Pool, it resets. 

       Hopefully someone will have some suggestions as to what processes are causing the time to increase.  I know in the short term I can lower the time from 20 minutes, but I would really like to try and figure out what is causing the delays.

       mike

      • #18349

        More information…

         

        After looking at this some more, it seems that the delay is in the response part of a Req/Res port.  In my Orch, I have parallel processes, one sends the response using a removeBOM pipeline, and another just does a file drop onto a local drive using the XML Transmit pipeline. 

         

        In Hat the time between the response and the file drop is what grows.  I think the time in HAT would be start time, but again I am guessing. 

         

        What would make the time between the kickoff of the file drop increase over time, and why would recycling the App Pool rest the tim?

        • #18359

          After all it was the fact that the SQL Server Agent was not started.  Not sure why this was not set to automatic, but now it is up.  Thing are flying now, and I am sure this is going to prevent other future issues such as a huge DTADb. 

           

          mike

Viewing 1 reply thread
  • The forum ‘BizTalk 2004 – BizTalk 2010’ is closed to new topics and replies.