0 Replies Latest reply on Jan 28, 2009 6:24 PM by Francis Limousy

    Load test using JMeter and Badboy

    Francis Limousy Newbie


      We have an ERP application developped using Seam that we're currently testing.
      I developped a simple test case using Badboy, and converted it to JMeter in order to run it on both applications.

      I notice that response times are quite longer in badboy than in JMeter (simulating a single user) and I was thinking: if JMeter isn't a browser and does not interpret the javascript (that's what the doc says), the response time can be far from the user feeling in front of his computer.
      In a more general way, pages really 2.0, containing a lot of richfaces elements, so a lot of javascript (drop-down menus, toggle panels, etc ...) are quite hard to test.

      In a way, the JMeter point of view is quite interesting; it can be interpreted as the total processing time, since request sending, to its comeback to the client computer.
      But I can't think of a way to simulate the browser processing time, aside adding an arbitrary thinking time (say, 2 sec maybe) for every request.
      Well I can't simulate a user resent in the middle of a high load usage of the server.
      I can launch the Badboy script in the middle of the JMeter load test but it's not really accurate and scriptable.

      Well did anybody managed to find a way to monitor the whole processing time (browser included) for many users ?