DX NetOps

Ranking OC-Console/Workstation to OC-Server performance

  • 1.  Ranking OC-Console/Workstation to OC-Server performance

    Broadcom Employee
    Posted Jul 13, 2015 11:50 AM

    Dear all,


    the CA Spectrum OC-Console covers an additional Java based debug application - which you can launch via "OC-Console" / "Help"-Menu - and here find

    "Debug Console" (or run via OC-weblogon / Administration / Debugging / Start Client Debug Console).

     

    Once this is open find "Console Test" tab - which allows to set on Top the number of test-cycles per test_0, test_1 and test_2.

    So - entering 10,5,10 into the top parameter input dialog and hitting the "Run Test" button will run test_0 = 10 times, test_1 = 5 and test_2 = 10 times in sequence. The output is saved to the dialog and you may save this.

     

    Background and value from these tests are:

    - test_0 allows to categorize the native network tcp session establishment from the Workstation to the OC-server host.

    - test_1 allows to categorize the OC-server to workstation data streaming / downstream performance

    - test_2 allows to categorize the "interactive" performance from workstation to the OC-server.

     

    Value(ms)test_0test_1test_2
    top< 6< 200< 6
    average/medium15 - 30500 - 400020 - 60
    critical> 200> 8000> 300

     

    Note - these are best-practice values which may require more detailled analysis in case of "trying to tune the performance". Find the the OC-Server configuration to run without "java.compiler=NONE" to improve test_1 a lot. But find also, that network topology (most likely affecting test_0) with number of hops for the traffic, also the workload conditions at intermediate hops and the workstation performance itself (i.e. jammed WLAN or connected via weak internet-provider / VPN-tunnel technology) can affect all values/tests. It is worth to test i.e. from a workstation which is part of the same network like the OC-server - and to compare the values per network segment and per route / different routes.

     

    Overall - you can make use of this test also during "non-office time" - and during "very critical workloaded network times" to see, if that directly affects the OC-Console performance when connecting and testing the communication to the OC-server. Also I would encourage you to test and document (find export option) these values at monthly level to find out about "changes" affecting the OC-Console performance.

     

     

    console_test.PNG

     

     

    Cheers, Joerg