Discussions

News: ECperf launches homepage on TheServerSide.com

  1. ECperf is a Specification and implementation of a benchmark designed to measure performance and scalability of J2EE servers. ECperf was prototyped and built in conjunction with J2EE server vendors, under the Java Community Process. As J2EE server benchmark results come in, they will be posted on ecperf.theserverside.com and announced on TheServerSide's homepage.

    What makes ECperf different from other benchmark initiatives (such as TPC) is that it includes a standard testing application complex enough to simulate a real business application. All benchmarks must be performed with this app (without any code changes), and final results/methodology must be approved by the ECperf committee before being released as official.

    Hopefully, ECperf should end all these childish cycles of performance comparisons released by IBM, BEA and Oracle. As the main accepted benchmark, whose results are posted on the trusted TheServerSide J2EE community, ECperf should hopefully provide definitive meaningful benchmark comparisons of todays servers.

    TheServerSide.com was chosen as the source for the ECperf home/results page due to its J2EE focus, vendor neutrality, credibility, and strong presence in the J2EE developer community.

    Check out the ECperf results page at ecperf.theserverside.com.
  2. Come on J2EE guys! Measure - show off! I can't wait to see the results...
  3. It's not launched yet - page says "August 1, 2001 - ECPerf Homepage launched on TheServerSide.com J2EE Community.".
  4. You are right Dimitri, although there will not be anything new appearing on August 1st. :) The ETA for the first result is not yet known.

    Floyd
  5. The page describes this as a J2EE benchmark. To be more specific, this is an EJB benchmark.
  6. I am always in favor of more information, so at first glance ecPerf is a welcome addition. It will allow comparisons that are not possible today. But why should a benchmark be limited to a particular implementation technology, as ecperf is limited to EJB?

    Why isn't TPC-W an appropriate benchmark? Since TPC-W is established, why are the J2EE vendors not supporting TPC-W with benchmark entries?

    In this light, establishing and supporting an EJB-only benchmark like ecPerf appears to limit information flow, A Bad Thing.

    ?
  7. Since BEA has never (to my recollection) shown up at any benchmark hosted by a im-partial third-party, despite numerous invitations from trade journals/magazines from the industry, I'm anxious to see if BEA actually shows up for this!

    Jason Weiss
    Sr. Software Engineer
    eBusiness Systems Group
    Sybase, Inc.
  8. BEA is and has been an active participant in independent benchmarking, where appropriate. BEA continues to be heavily involved with ecPerf on an ongoing basis.

    Tyler
  9. RE: the Sybase guy's assertion that BEA had never "shown up."

    The guys at OrionServer.com (now part of Oracle's OAS -- actually OC4J) up to six months ago published the results of their benchmarks of ALL app servers - Weblogic was one of them, as was IIS/ASP vs. JSP (they used to also include straight CGI, mod_perl, Python/Zope and PHP). As noted on their current Web page, they ceased to include Weblogic when BEA threatened suit against the two guys who wrote OrionServer. I Googled for a result and here it is (until BEA orders it gone)

    http://www2.real-time.com/rte-tomcat/2000/Jul/msg01759.html

    As you can see Weblogic, in an independent, fairly balanced test isn't so fast after all.

    Here's OrionServer's current benchmarks. Read 'em and weep (or better yet, download the tests and do'em yourselves. Orion, Tomcat, JBoss, PHP, ZOPE, etc. are all freely downloadable - and now we have another third party test - that customers won't be able to publish the results of - ECperf - just WHAT is BEA hiding, OTHER THAN THEIR OUTRAGEOUS PRICE TAG?) Orionserver benchmark page:

    http://orionserver/benchmarks/benchmark.html

    Michael J. Cannon
    mcannon@ubiquicomm.com