ECperf performance benchmark kit update 1 posted


News: ECperf performance benchmark kit update 1 posted

  1. The ECperf kit, which also includes a Specification, is a benchmark designed to measure performance and scalability of J2EE servers. The kit includes a non-trivial EJB/Servlet/JSP application. Results published after October 11, 2001 must use this new update. Ecperf 1.0 has around since June, but unfortunately, no vendors have published results yet.

    Check out ECPerf Update 1.

    Check out the official ECperf Results Page on TheServerSide.

    Threaded Messages (14)

  2. ECperf appropriate?[ Go to top ]


    I am doing an application server study for my thesis.
    The purpose is to do an in depth review of 2 application servers. The Reviews section has already helped me a lot in preselecting these 2 servers.

    On I read that ECperf is a tool for both server vendor and J2EE user. But how much effort (time to learn and use) is required to compare 2 servers with ECperf? I was wondering if it will leave me enough time to discuss other issues than scalability and performance (say IDE integration, web services support etc.)

    Can anybody tell me about his experience?

  3. ECperf appropriate?[ Go to top ]


    I am running the ECperf benchmark on WebSphere, WebLogic and JBoss. The project length is 5 weeks and this includes setting up and configuring the 3 servers on W2k and Linux. Creating the deployment descriptors, running the tests, performance tuning and writing up the documentation on how we did it, any installation issues and the results.

    So in short I would say you have time to run ECperf against 2 application servers. I would also recommend that as part of you review some custom tests that test the throughput of Entity and Session beans where there is no database access as this will show you the overhead of the application servers implementation of the EJB & J2EE semantics.


    P.S. I will not be able to publically publish the results due to licensing restrictions
  4. ECperf appropriate?[ Go to top ]

    Thanks a lot Paul!

    You've given me a "GO" for ECperf.

    One more question: how did you select the Entity Bean methods with no database access? Is this completely predictable considering activation/passivation?

    Thanks again,
  5. ECperf appropriate?[ Go to top ]


    What I was suggesting was that you create your own BMP Entity bean trst suite where the beans do not load/store information to the database and just cache it in instance variables. As what you would be testing is the time to go through the stack from the client test through the security and transaction code to your method call it would not be necessary to have the actual data in the entity preserved through activation and passivation.

    For ECperf you will of course need to run it with the database calls as there are strict rules about what you can and can't modify.

  6. ECperf appropriate?[ Go to top ]

    Ok Paul, you're right: with BMP you can eliminate the DB.

    I suggest using JUnit or and extension of it, like Cactus?

  7. ECperf appropriate?[ Go to top ]


    > P.S. I will not be able to publically publish the results due to licensing restrictions...

    I think that says it all why the ECperf is appropriate or
    sigificant! Sun is saying here is a test, see what
    performance numbers you get. You decide what they mean.

    EULA's restrict you in telling any one what you
    find. The vendors do *not* want uncontrolled or
    unsolicited press in this situation.

    Benchmarks and standards. You got to love and support them,
    after all there are so may to choose from...



    P.S. Statistics show that most informed individuals
    believe that 50% of statistics are statistically
    insignificant in there decision making, 50% of the

  8. ECperf appropriate?[ Go to top ]

    Paul, can you please mail me at pietervangorp at skynet dot be please? I've got some more questions about your project with ECperf.

    Thanks in advance,
  9. Ecperf 1.0 has around since June, but unfortunately, no vendors have published results yet.

    I'm curious why - does anybody have any theories? There anecdotical are things like thia BEA claim and Oracle response, but no ECPerf results.
  10. I guess every vendor is looking for the other to publish the results first.
  11. That's why I wanted to do it myself... but can anybody tell me how much effort this takes? Seems a lot to me, because otherwise there already would have been a lot of others before me, right?
  12. Borland just anounced they have released their ECperf benchmark kit.

    Borland Takes Lead in Industry Standard Benchmarking of J2EE™ Application Servers
  13. Is it possible to run ECPerf against a cluster?
  14. I'll have to stop talking to myself... but it is possible to run ECperf against a cluster :)
  15. yes.