SPECJAppServer2002: IBM leads multinode category with new result

Discussions

News: SPECJAppServer2002: IBM leads multinode category with new result

  1. IBM submitted a new SpecjAppServer2002 benchmark number running WebSphere 5.1 on an eServer xSeries x335, in the multinode category.

    IBM now leads the multinode category in performance with 2,575.34 total operations per second (TOPS) with a fairly low submission on price/performance ($330.07/TOPS).

    Configuration:

    Application Server: WebSphere 5.1 Application Server, with 2 Intel Xeon DP CPUs at 3200 Mhz
    Hardware: eServer xSeries 335 Cluster
    JVM: IBM's 1.4.1 JRE
    JDBC Driver: IBM DB2 JDBC Universal Driver Provider (Type 2)
    OS: SuSE Linux Enterprise Server 8 SP2A with ReiserFS

    Database: DB2 Universal Database v8.1.1.32 FP4, Enterprise Server Edition running on AIX

    There were 9 application servers systems running, with 1 database.

    You can view the details on this result at:
    http://www.spec.org/osg/jAppServer2002/results/res2003q4/jAppServer2002-20031111-00020.html

    You can view all the results at:
    http://www.spec.org/osg/jAppServer2002/results/jAppServer2002.html
  2. ...by quite a bit:

    HP
    HP Integrity Superdome / BEA WebLogic Server 8.1 SP1 on HP-UX 11i v2
    4496.28
    652.00 (US$)
    MultipleNode
    Text  HTML 

    HP
    HP rp7410 and HP Superdome / BEA WebLogic Server 8.1 on HP-UX 11i
    4158.13
    1175.16 (US$)
    MultipleNode
    Text  HTML 
  3. Right! but...

    HP J2EE Application Server (7 systems)
      # of CPUs: 8
      Memory (MB): 32768

    IBM J2EE Application Server (9 systems)
      # of CPUs: 2
      Memory (MB): 3072

    HP Database Server (1 system)
      # of CPUs: 40
      Memory (MB): 40960

    IBM Database Server (1 system)
      # of CPUs: 8
      Memory (MB): 16GB

    HP Total System Cost: US$ 4,886,443
    IBM Total System Cost: US$ 850,025

    Ciao, D.
  4. Frankly spoken, I don't give damn about any of these benchmarks, since they are absolutely meaningsless (de do do do, de da da da ...) and only serve marketing purposes.

    Performance is _VERY_ heavy influenced by the application and not only by the system environment. So, in the end, what does some kind of benchmark tell you? Only that the application tested by the benchmark, using a certain system environment, has <xyz> TOPS. Unfortunately I usually develop a different application (and even more: deploy it on a different system environment).

    Regards,
        Dirk