RE: Re: Cayenne performance testing

From: Gentry, Michael \(Contractor\) ("Gentry,)
Date: Wed Feb 01 2006 - 09:15:50 EST

  • Next message: Ahmed Mohombe: "Re: Cayenne performance testing"

    I kind of disagree about the web application testing part. I think if
    we are going to profile code, it should be Cayenne code. Let the
    Tapestry/Struts/JSP/JSF/Tomcat/Jetty/Resin/WebSphere/JBoss/etc guys
    worry about profiling their code.

    As for test suites, I know PostgreSQL ships with pgbench, which is a
    crude benchmark. Perhaps it could be re-written as a Cayenne
    application and then you can compare raw C access versus ORM, too. I
    expect the raw C version to be faster, but ORM is almost always worth a
    little performance decrease.

    /dev/mrg

    -----Original Message-----
    From: news [mailto:new..ea.gmane.org] On Behalf Of Ahmed Mohombe
    Sent: Wednesday, February 01, 2006 5:15 AM
    To: cayenne-deve..bjectstyle.org
    Subject: Re: Cayenne performance testing

    > So I was wondering what would people recommend as far as such testing
    > setup. Is it worth looking at some other tools beyond Ant and Junit,
    > like for instance JMeter?
    IMHO yes. So far for free solutions, jMeter it seems the best option.
    It's very easy to use, but the test scenarios must be well designed to
    make sense.

    > We are not testing a web application after all...
    It doesn't matter. Most of the use scenarios of Cayenne are in a
    webapplication.
    In your case, just minimize the webapplication overhead as much as
    possible (to just
    call Cayenne code).
    If you make performance tests that call cayenne code directly without
    any web tier, they won't be
    that realistic, and maybe not even 100% the same API will be used(at
    least not in the same time
    sequence distribution).

    With the Click framework, we want the same thing, but we need a complete
    (automated) test
    coverage with Selenium first. Only after that can we take JMeter. Of
    course, it would be nice
    if the same application would be usable for both, but I'm not that sure.
    Anyway you proposed that JPetStore as base test application. I'm still
    analyzing it, but so far for
      coverage automated test looks OK.
    So far, I can't judge if it's good for performance tests, so maybe you
    could also check if that
    would be a good test case for the performance tests.

    Ahmed.



    This archive was generated by hypermail 2.0.0 : Wed Feb 01 2006 - 09:15:53 EST