VPS L out preforms VPS Triple X

Discussion in 'Linux VPS/Dedicated - General' started by klurt, Sep 15, 2007.

  1. klurt

    klurt New Member

    Hi

    I just conducted a benchmark between Knownhot VPS L and VPS Triple X and the results are disturbing.

    Both VPSes are identically configured in regard to services running, packages installed, and control panels (cPanel). The only difference between the two is that the VPS L is running a FTP server while the Triple X is not.

    Using unixbench-4.1.0

    VPS L

    BYTE UNIX Benchmarks (Version 4.1-wht.1)
    System -- Linux 2.6.9-023stab044.4-entnosplit #1 SMP Thu May 24 17:58:35 MSD 2007 i686 athlon i386 GNU/Linux
    /dev/vzfs 20971520 5778232 15193288 28% /

    Start Benchmark Run: Sat Sep 15 14:18:32 PDT 2007
    14:18:32 up 20 days, 15:17, 1 user, load average: 0.06, 0.01, 0.00

    End Benchmark Run: Sat Sep 15 14:29:54 PDT 2007
    14:29:54 up 20 days, 15:28, 1 user, load average: 13.48, 6.00, 2.74


    INDEX VALUES
    TEST BASELINE RESULT INDEX

    Dhrystone 2 using register variables 376783.7 7180826.4 190.6
    Double-Precision Whetstone 83.1 1419.8 170.9
    Execl Throughput 188.3 1433.2 76.1
    File Copy 1024 bufsize 2000 maxblocks 2672.0 32994.0 123.5
    File Copy 256 bufsize 500 maxblocks 1077.0 7177.0 66.6
    File Read 4096 bufsize 8000 maxblocks 15382.0 282726.0 183.8
    Pipe Throughput 111814.6 768425.9 68.7
    Pipe-based Context Switching 15448.6 191483.3 123.9
    Process Creation 569.3 5088.1 89.4
    Shell Scripts (8 concurrent) 44.8 363.6 81.2
    System Call Overhead 114433.5 1031362.5 90.1
    =========
    FINAL SCORE 106.9

    =======================================

    VPS Triple X


    BYTE UNIX Benchmarks (Version 4.1-wht.1)
    System -- Linux 2.6.9-023stab040.1-entnosplit #1 SMP Mon Jan 15 23:19:03 MSK 2007 i686 athlon i386 GNU/Linux
    /dev/vzfs 31457280 5611632 25845648 18% /

    Start Benchmark Run: Sat Sep 15 17:33:25 EDT 2007
    17:33:25 up 76 days, 20:31, 1 user, load average: 0.00, 0.00, 0.10

    End Benchmark Run: Sat Sep 15 17:44:35 EDT 2007
    17:44:35 up 76 days, 20:42, 1 user, load average: 13.78, 5.90, 2.70


    INDEX VALUES
    TEST BASELINE RESULT INDEX

    Dhrystone 2 using register variables 376783.7 3492368.2 92.7
    Double-Precision Whetstone 83.1 674.5 81.2
    Execl Throughput 188.3 1554.3 82.5
    File Copy 1024 bufsize 2000 maxblocks 2672.0 71735.0 268.5
    File Copy 256 bufsize 500 maxblocks 1077.0 35535.0 329.9
    File Read 4096 bufsize 8000 maxblocks 15382.0 316303.0 205.6
    Pipe Throughput 111814.6 675453.4 60.4
    Process Creation 569.3 3782.4 66.4
    Shell Scripts (8 concurrent) 44.8 257.9 57.6
    System Call Overhead 114433.5 379411.0 33.2
    =========
    FINAL SCORE 98.6


    While there is not a large difference between the two scores, I can't help but feel that I'm paying an extra $45 a month for a VPS Triple X when I could just stick with the VPS L and get the same performance. Am I missing something here?
     
  2. Jleagle

    Jleagle New Member

    Are you not paying that extra 45 bucks for the extra ram, bandwidth, hard drive etc?
     
  3. KH-Paul

    KH-Paul CTO Staff Member

    klurt,

    All VPSs are hosted in mixed enviroment on physical nodes. This means that physical server run a mix of M's, L's, ..., Triple X's. With Triple X you have higher limits for all VPS related parameters but it doesn't (and shouldn't) give you any better pripority for CPU time and/or Disk I/O.

    Regards,
    Paul
     
  4. ppc

    ppc Moderator

    Klurt,

    The first thing you have to notice is that all of the KH VPS plans say CPU is equal share which means as Paul said no matter what plan you get, each will still have equal share of the CPU.

    As to why your "L" plan performed better on that test than on your "XXL" test is as follows:

    I assume their on different nodes. The node on which your "L" VPS was on might have had less other server activity than the node that your "XXL" VPS had at the time you started the test.

    Your really are paying more for additional bandwidth, hard drive space and memory which I believe wouldnt really effect the outcome of the benchmark test to that much of an extent.

    I hope this clears things up.

    Regards,
     
  5. ppc

    ppc Moderator

    One more thing: how do you know your VPS "L" and VPS "XXL" were doing the same things behind the scenes.

    Meaning maybe your "XXL" vps was updating cpanel or its spam definitions or some other proccess which your "L" was not.
     

Share This Page