Page 1 of 21 1234 11 ... LastLast
Results 1 to 40 of 816
  1. #1
    Join Date
    Jun 2004
    Location
    Sao Paulo/ Brasil
    Posts
    135

    VPS Benchmarks - Post Your vps performance

    Download: UnixBench v4.1.0 - WHT Variant

    # wget http://members.dslextreme.com/users/...1.0-wht.tar.gz
    # gunzip -dvc unixbench-4.1.0-wht.tar.gz | tar xvf -
    # cd unixbench-4.1.0-wht
    # make
    # ./Run

    Please download and run it on your VPS. Then post your result here for discussion. Thanks!
    Last edited by Mike V; 02-23-2009 at 04:59 PM. Reason: Inserted code tag
      0 Not allowed!

  2. #2
    Join Date
    Jun 2004
    Location
    Sao Paulo/ Brasil
    Posts
    135
    Code:
    powervps 
    Powervps.com
    
    CPANEL POWER-1
    $44.50 per month 
    
    RedHat Linux
    Equal share CPU
    cPanel/WHM (latest edition)
    Fantastico De-Luxe
    Clientexec
    1GB Burst RAM
    256MB RAM guaranteed
    10GB Disk Space
    150GB Monthly Transfer 
    
    ==============================================================
    BYTE UNIX Benchmarks (Version 4.1-wht)
    System -- Linux host.boxhost.com.br 2.4.20-021stab028.7.777-enterprise #1 SMP Thu Apr 14 13:59:38 MSD 2005 i6
    86 athlon i386 GNU/Linux
    /dev/vzfs             13762560   3427371  10335189  25% /
    
    Start Benchmark Run: Sat May 28 16:10:58 EDT 2005
     16:10:58 up 7 days, 22:21,  1 user,  load average: 0.11, 0.05, 0.01
    
    End Benchmark Run: Sat May 28 16:22:22 EDT 2005
     16:22:22 up 7 days, 22:32,  1 user,  load average: 16.05, 6.30, 2.81
    
    INDEX VALUES 
    TEST BASELINE RESULT INDEX
    
    Dhrystone 2 using register variables 376783.7 2429554.3 64.5
    Double-Precision Whetstone 83.1 453.3 54.5
    Execl Throughput 188.3 1269.7 67.4
    File Copy 1024 bufsize 2000 maxblocks 2672.0 11129.0 41.7
    File Copy 256 bufsize 500 maxblocks 1077.0 4529.0 42.1
    File Read 4096 bufsize 8000 maxblocks 15382.0 197876.0 128.6
    Pipe Throughput 111814.6 200931.8 18.0
    Pipe-based Context Switching 15448.6 55830.3 36.1
    Process Creation 569.3 2345.2 41.2
    Shell Scripts (8 concurrent) 44.8 220.9 49.3
    System Call Overhead 114433.5 133776.5 11.7
    =========
    FINAL SCORE 42.5
      0 Not allowed!

  3. #3
    Join Date
    Oct 2003
    Location
    Chattanooga
    Posts
    8,985
    This trustable?
    If so why the different versioning rather than just the default UnixBench?
    David
    Web hosting by Fused For businesses with more important things to do than worry about their hosting.
      0 Not allowed!

  4. #4
    Join Date
    Jun 2004
    Location
    Sao Paulo/ Brasil
    Posts
    135
    i find it in post vps vs dedicated in dedicated forum , i think its trustable.
      0 Not allowed!

  5. #5
    Join Date
    Jan 2004
    Location
    North Yorkshire, UK
    Posts
    4,163
      0 Not allowed!

  6. #6
    Join Date
    Jun 2000
    Location
    Washington, USA
    Posts
    5,991
    RH9 VPS Mini(75Mhz CPU with burst, and 96MB RAM with burst)

    Code:
                         INDEX VALUES
    TEST                                        BASELINE     RESULT      INDEX
    
    Dhrystone 2 using register variables        376783.7  7721598.3      204.9
    Double-Precision Whetstone                      83.1      563.8       67.8
    Execl Throughput                               188.3     2040.9      108.4
    File Copy 1024 bufsize 2000 maxblocks         2672.0    18064.0       67.6
    File Copy 256 bufsize 500 maxblocks           1077.0     4546.0       42.2
    File Read 4096 bufsize 8000 maxblocks        15382.0   315920.0      205.4
    Pipe Throughput                             111814.6   308461.1       27.6
    Pipe-based Context Switching                 15448.6    86917.4       56.3
    Process Creation                               569.3     5186.0       91.1
    Shell Scripts (8 concurrent)                    44.8      518.9      115.8
    System Call Overhead                        114433.5   144991.7       12.7
                                                                     =========
         FINAL SCORE                                                      69.5
      0 Not allowed!

  7. #7
    Join Date
    Nov 2003
    Location
    Canada
    Posts
    881
    A 196MB VPS on XEN technology.

    Code:
    BYTE UNIX Benchmarks (Version 4.1-wht)
    System -- Linux vps36156.u2knetworks.com 2.6.11.10-xenU #1 Sun May 22 11:42:16 BST 2005 i686 GNU/Linux
    /dev/sda1             10321208   1190972   8605948  13% /
    
    Start Benchmark Run: Sat May 28 23:54:22 CDT 2005
     23:54:22 up 1 day, 23:11,  1 user,  load average: 0.05, 0.01, 0.00
    
    End Benchmark Run: Sun May 29 00:05:12 CDT 2005
     00:05:12 up 1 day, 23:21,  1 user,  load average: 10.60, 5.00, 2.33
    
    
                         INDEX VALUES
    TEST                                        BASELINE     RESULT      INDEX
    
    Dhrystone 2 using register variables        376783.7  4476396.8      118.8
    Execl Throughput                               188.3     2131.1      113.2
    File Copy 1024 bufsize 2000 maxblocks         2672.0    45308.0      169.6
    File Copy 256 bufsize 500 maxblocks           1077.0    27593.0      256.2
    File Read 4096 bufsize 8000 maxblocks        15382.0   123794.0       80.5
    Pipe Throughput                             111814.6  1102043.3       98.6
    Pipe-based Context Switching                 15448.6   142662.4       92.3
    Process Creation                               569.3     2395.9       42.1
    Shell Scripts (8 concurrent)                    44.8      294.9       65.8
    System Call Overhead                        114433.5  1130399.6       98.8
                                                                     =========
         FINAL SCORE                                                     101.6
      0 Not allowed!

  8. #8
    PowerVPS Cpanel Power 1 plan

    INDEX VALUES
    TEST BASELINE RESULT INDEX

    Dhrystone 2 using register variables 376783.7 4081911.9 108.3
    Double-Precision Whetstone 83.1 1095.7 131.9
    Execl Throughput 188.3 843.1 44.8
    File Copy 1024 bufsize 2000 maxblocks 2672.0 6922.0 25.9
    File Copy 256 bufsize 500 maxblocks 1077.0 2079.0 19.3
    File Read 4096 bufsize 8000 maxblocks 15382.0 118911.0 77.3
    Pipe Throughput 111814.6 167327.9 15.0
    Pipe-based Context Switching 15448.6 32251.2 20.9
    Process Creation 569.3 1180.6 20.7
    Shell Scripts (8 concurrent) 44.8 199.9 44.6
    System Call Overhead 114433.5 252372.6 22.1
    =========
    FINAL SCORE 36.6



      0 Not allowed!

  9. #9
    For the newbies, would someone explain what a good score would be?
      0 Not allowed!

  10. #10
    Join Date
    May 2004
    Posts
    1,663
    The higher the score, the better. For a VPS, anything above 50 is good, even in the 30's is ok. I'm not an expert, that's just what I think.

    By the way, I think PowerVPS has been restricting us too greatly lately.
      0 Not allowed!

  11. #11
    Join Date
    Jun 2004
    Location
    Sao Paulo/ Brasil
    Posts
    135
    see this result,

    in a 196 mb vps, the benchmark result 101 is not a real result.
    in 96 mb vps 70 in final score is real result ?
    its unreal.

    this server have only two vps ?



    2x Intel(R) Xeon(TM) CPU 2.80GHz with HT enabled.
    2GB memory


    code:--------------------------------------------------------------------------------
    ==============================================================
    BYTE UNIX Benchmarks (Version 4.1-wht)
    System -- Linux tornado.he.net 2.4.25-ow1 #6 SMP Thu Dec 16 10:14:02 PST 2004 i6
    86 unknown
    /dev/hda1 198980892 79398832 119582060 40% /

    Start Benchmark Run: Mon Apr 11 15:47:46 PDT 2005
    15:47:46 up 56 days, 4:23, 3 users, load average: 0.75, 0.51, 0.42

    End Benchmark Run: Mon Apr 11 15:58:19 PDT 2005
    15:58:19 up 56 days, 4:34, 2 users, load average: 15.50, 6.67, 3.16


    INDEX VALUES
    TEST BASELINE RESULT INDEX

    Dhrystone 2 using register variables 376783.7 8232074.0 218.5
    Double-Precision Whetstone 83.1 595.1 71.6
    Execl Throughput 188.3 5723.6 304.0
    File Copy 1024 bufsize 2000 maxblocks 2672.0 87670.0 328.1
    File Copy 256 bufsize 500 maxblocks 1077.0 27113.0 251.7
    File Read 4096 bufsize 8000 maxblocks 15382.0 437683.0 284.5
    Pipe Throughput 111814.6 1272487.7 113.8
    Pipe-based Context Switching 15448.6 206647.6 133.8
    Process Creation 569.3 12918.9 226.9
    Shell Scripts (8 concurrent) 44.8 864.0 192.9
    System Call Overhead 114433.5 940977.9 82.2
    =========
    FINAL SCORE 179.2




    Originally posted by Lev
    The higher the score, the better. For a VPS, anything above 50 is good, even in the 30's is ok. I'm not an expert, that's just what I think.

    By the way, I think PowerVPS has been restricting us too greatly lately.
      0 Not allowed!

  12. #12
    Cool tool, thanks for sharing.
      0 Not allowed!

  13. #13
    Join Date
    Jun 2000
    Location
    Washington, USA
    Posts
    5,991
    in 96 mb vps 70 in final score is real result ?
    It's a real result, from an active VPS. Our nodes are kept at light load, and we only use highend server hardware. E.g. all SCSI drives, etc.
      0 Not allowed!

  14. #14
    ok so its 96mb ram, what kind of cpu useage allowance does that have? If I buy that plan from you will you personally guarantee I can run that test anytime day or night and get within 10% of that result? Past a certain minimum the level of ram doesnt matter that much, its cpu power, what cpu level your allowed to burst to.
    --
    Rich
      0 Not allowed!

  15. #15
    Join Date
    Apr 2005
    Location
    Oz
    Posts
    3,498
    Does anybody have benchmarks for the Steadfast Networks Bronze VPS (with Plesk if possible).
    If you do please post them for me!

    Thanks,
      0 Not allowed!

  16. #16
    UnixBench doesn't need RAM.

    You could have a 32MB machine and it would give a similar result to a 512MB machine.

    We've tried this ourselves, this is true.


    In any event, a good score is anything 30 and up.

    As a reference, I have an IBM ThinkPad A22 notebook I'm using right now, with a P3/900 & 512MB, and it generated a 61 on UnixBench; compiled under CentOS 4.

    I'll throw in another reference. Totally unloaded Dell PowerEdge 2850 w/Dual Xeon 2.8 EM64T, generated a 198 on the same test; compiled under CentOS 4 as well.
    EuroVPS VPS Hosting - Virtual Private Servers | Web Hosting | Dedicated Servers
    Providing Reliable Plesk and cPanel Servers since 2004, now offering low priced Xen & VMware VPS in Amsterdam
    UK +44.203.355.6681 / Amsterdam +31.208.202.120
      0 Not allowed!

  17. #17
    Join Date
    Jun 2000
    Location
    Washington, USA
    Posts
    5,991
    Originally posted by richy
    ok so its 96mb ram, what kind of cpu useage allowance does that have? If I buy that plan from you will you personally guarantee I can run that test anytime day or night and get within 10% of that result? Past a certain minimum the level of ram doesnt matter that much, its cpu power, what cpu level your allowed to burst to.
    That VPS was set to a 75Mhz guarantee, with ability to burst to full server capacity. Virtuozzo allows bursting of resources when they aren't being used.

    I can't guarantee what result someone would get, we have various nodes of different configuration. However, in previous benchmarks our customers have posted pretty nice scores.
      0 Not allowed!

  18. #18
    So basically you could have picked you least loaded node, wouldn't it be a much more honest representation of your product for a customer on a random node which is more likely to be at the real world utilisation level to post results?

    Having all this been said regarding speed, its also important to mention of course the requirement to check other factors when buying a vps, not just network but also the quality of support and the longevity of the company. Not all startups will fail, but your safer with a company that has been running successfully for a respectable amount of time without major issues. Less likely to tank or sellout

    In a fair test servint scores about 13 on their lowest account, this is because they cap the amount of cpu you can burst to stop you abusing the server or causing performance issues for other users. They've been around just a bit longer then most of the competition (over 10 years), I know who I put my faith in It would be nice if they tested out raising that cap a little but theyre happy their product is stable and good value as it is, they know the business better then most, so I think Ill go with stability rather then unreliable burstability from a company that could tank or sellout.
    --
    Rich
      0 Not allowed!

  19. #19
    Join Date
    Apr 2002
    Location
    Orange County, CA
    Posts
    338
    The baseline score (10.0) for this benchmark was obtained on a Pentium 233MMX machine with 128MB RAM and 7200rpm SCSI hard disk running FreeBSD 2.x. Thus in theory, a P4 2.4GHz should get a score of 100.0 if all components are ten times faster.

    To properly evaluate a VPS, you should run the benchmark under several conditions:

    (1) Temporarily stop all non-essential services (Apache, MySQL, control panel, DNS, mail server, and cron scripts), the resulting benchmark score will tell you how much CPU power is allocated to you by the VPS company. If this "idle score" is low during the day but much higher at night, then the VPS company probably put too many VPS's on the same machine.

    (2) Once you have the VPS running the normal load of services, do the benchmark again at different times of the day. If your VPS doesn't have enough memory (less than 256MB) to sufficiently run your web sites, then you will likely notice an immediate drop in benchmark score. Consider upgrading your VPS in that case.

    One shouldn't choose a VPS solely on good benchmark scores. Other factors such as support, network connectivity, server reliability, uptime, and VPS features are also important. ServInt provides superb support but minimum CPU power, which is only suited for small, light traffic web sites in my opinion.
      0 Not allowed!

  20. #20
    I agree however I cant say what their higher packages would score. Im not actually sure if what we are seeing here is a difference in the basic cpu power availible or a difference in the burstability. Obviously some providers dont limit the burst as much as servint, and given servints reputation for quality and technical knowledge I would tend to think they have got it right
    --
    Rich
      0 Not allowed!

  21. #21

    my score

    Code:
    LiquidWeb
    
    VPS Plan 1
    60$ a month
    
    AMD Dual Opteron Server
    256MB RAM Guaranteed
    cPanel/WHM (latest edition)
    Fantastico De-Luxe
    1GB Burstable RAM
    20GB Disk Space
    200GB Monthly Transfer 
    
    BYTE UNIX Benchmarks (Version 4.1-wht)
    System -- Linux
    2.4.20-021stab028.7.777-smp #1 SMP
    Thu Apr 14 15:04:58 MSD 2005 i686 athlon i386 GNU/Linux
    
    /dev/vzfs             20971520   3270980  17700540  16% /
    
    Start Benchmark Run: Fri Jun  3 03:06:15 MDT 2005
     03:06:15 up 4 days,  9:44,  1 user,  load average: 1.34, 1.03, 0.98
    
    End Benchmark Run: Fri Jun  3 03:19:47 MDT 2005
     03:19:47 up 4 days,  9:58,  1 user,  load average: 15.08, 6.57, 3.51
    
    
                         INDEX VALUES            
    TEST                                        BASELINE     RESULT      INDEX
    
    Dhrystone 2 using register variables        376783.7  4272842.5      113.4
    Double-Precision Whetstone                      83.1      934.6      112.5
    Execl Throughput                               188.3     1269.4       67.4
    File Copy 1024 bufsize 2000 maxblocks         2672.0    18006.0       67.4
    File Copy 256 bufsize 500 maxblocks           1077.0     5599.0       52.0
    File Read 4096 bufsize 8000 maxblocks        15382.0   193801.0      126.0
    Pipe Throughput                             111814.6   370659.3       33.1
    Pipe-based Context Switching                 15448.6    59811.3       38.7
    Process Creation                               569.3     2081.7       36.6
    Shell Scripts (8 concurrent)                    44.8      241.0       53.8
    System Call Overhead                        114433.5   346119.4       30.2
                                                                     =========
         FINAL SCORE                                                      58.8
      0 Not allowed!

  22. #22
    Join Date
    Apr 2005
    Location
    Oz
    Posts
    3,498
    Here are my scores for my Steadfast Networks Bronze VPS with Plesk:

    Code:
    
    Start Benchmark Run: Wed Jun  8 18:34:15 CDT 2005
     18:34:15  up 1 day,  2:40,  1 user,  load average: 0.65, 0.25, 0.09
    
    End Benchmark Run: Wed Jun  8 18:47:12 CDT 2005
     18:47:12  up 1 day,  2:53,  1 user,  load average: 14.99, 6.34, 2.95
    
    
                         INDEX VALUES
    TEST                                        BASELINE     RESULT      INDEX
    
    Dhrystone 2 using register variables        376783.7  6997557.9      185.7
    Double-Precision Whetstone                      83.1      577.4       69.5
    Execl Throughput                               188.3     2424.6      128.8
    File Copy 1024 bufsize 2000 maxblocks         2672.0     7696.0       28.8
    File Copy 256 bufsize 500 maxblocks           1077.0     3853.0       35.8
    File Read 4096 bufsize 8000 maxblocks        15382.0   333018.0      216.5
    Pipe Throughput                             111814.6   300763.3       26.9
    Pipe-based Context Switching                 15448.6    89108.6       57.7
    Process Creation                               569.3     3364.1       59.1
    Shell Scripts (8 concurrent)                    44.8      508.9      113.6
    System Call Overhead                        114433.5   188844.5       16.5
                                                                     =========
         FINAL SCORE                                                      63.2
    
      0 Not allowed!

  23. #23
    Join Date
    Nov 2003
    Posts
    63
    Here are some tests I ran today on machines at Godaddy, PowerVPS, and Servint. I've run them twice on a couple machines with similar results, but the results may be different at night, etc. If there are significant differences, I'll try to edit this post.

    Godaddy VPS : About $51.00/month, 500GB bandwidth, 20GB disk, Plesk.
    Support takes several hours, is mediocre.

    Code:
    BYTE UNIX Benchmarks (Version 4.1-wht)
    ==============================================================
    BYTE UNIX Benchmarks (Version 4.1-wht)
    System -- Linux xxx.prod.phx1.secureserver.net 2.4.20-021stab022.10.777-enterprise #1 SMP Mon Jan 10 12:33:35 MSK 2005 i686 i686 i386 GNU/Linux
    /dev/vzfs             20971520    602416  20369104   3% /
    
    Start Benchmark Run: Mon Jun 27 15:16:10 MST 2005
     15:16:10  up 12 days,  7:06,  1 user,  load average: 0.17, 0.06, 0.01
    
    End Benchmark Run: Mon Jun 27 15:26:56 MST 2005
     15:26:56  up 12 days,  7:17,  1 user,  load average: 15.92, 6.44, 2.80
    
    
                         INDEX VALUES
    TEST                                        BASELINE     RESULT      INDEX
    
    Dhrystone 2 using register variables        376783.7  3730713.3       99.0
    Double-Precision Whetstone                      83.1      531.0       63.9
    Execl Throughput                               188.3     1063.8       56.5
    File Copy 1024 bufsize 2000 maxblocks         2672.0     2104.0        7.9
    File Copy 256 bufsize 500 maxblocks           1077.0     2055.0       19.1
    File Read 4096 bufsize 8000 maxblocks        15382.0   102427.0       66.6
    Pipe Throughput                             111814.6    75245.9        6.7
    Pipe-based Context Switching                 15448.6    28991.2       18.8
    Process Creation                               569.3      455.6        8.0
    Shell Scripts (8 concurrent)                    44.8      103.9       23.2
    System Call Overhead                        114433.5    54702.9        4.8
                                                                     =========
         FINAL SCORE                                                      21.3

    PowerVPS cPanel Power 2, $84.50/month. 350GB transfer, 20GB disk. Host provides nice backup. Support is prompt and friendly, but I haven't found a phone # for support.
    Code:
    BYTE UNIX Benchmarks (Version 4.1-wht)
    System -- Linux host.xxxx.com 2.4.20-021stab028.12.777-enterprise #1 SMP Fri May 13 15:50:02 MSD 2005 i686 athlon i386 GNU/Linux
    /dev/vzfs             19922944   2073921  17849023  11% /
    
    Start Benchmark Run: Mon Jun 27 17:46:30 EDT 2005
     17:46:30 up 5 days, 17:48,  1 user,  load average: 0.30, 0.27, 0.75
    
    End Benchmark Run: Mon Jun 27 17:58:18 EDT 2005
     17:58:18 up 5 days, 17:59,  1 user,  load average: 13.04, 5.81, 3.05
    
    
                         INDEX VALUES
    TEST                                        BASELINE     RESULT      INDEX
    
    Dhrystone 2 using register variables        376783.7  4279669.3      113.6
    Double-Precision Whetstone                      83.1      965.4      116.2
    Execl Throughput                               188.3      873.3       46.4
    File Copy 1024 bufsize 2000 maxblocks         2672.0     8680.0       32.5
    File Copy 256 bufsize 500 maxblocks           1077.0     2316.0       21.5
    File Read 4096 bufsize 8000 maxblocks        15382.0    75084.0       48.8
    Pipe Throughput                             111814.6   101502.8        9.1
    Pipe-based Context Switching                 15448.6    42556.3       27.5
    Process Creation                               569.3      942.7       16.6
    Shell Scripts (8 concurrent)                    44.8      155.7       34.8
    System Call Overhead                        114433.5   215741.1       18.9
                                                                     =========
         FINAL SCORE                                                      33.3

    Servint VPS 350GB transfer, 20GB disk, cPanel. Excellent support including phone support.

    Code:
    BYTE UNIX Benchmarks (Version 4.1-wht)
    System -- Linux vps.*********** 2.4.20-021stab028.12.777-enterprise #1 SMP Fri May 13 15:50:02 MSD 2005 i686 i686 i386 GNU/Linux
    /dev/vzfs             20971520   4235158  16736362  21% /
    
    Start Benchmark Run: Mon Jun 27 17:29:48 EDT 2005
     17:29:48  up 7 days,  5:27,  1 user,  load average: 0.24, 0.10, 0.08
    
    End Benchmark Run: Mon Jun 27 17:42:27 EDT 2005
     17:42:27  up 7 days,  5:39,  1 user,  load average: 14.99, 6.50, 3.07
    
    
                         INDEX VALUES
    TEST                                        BASELINE     RESULT      INDEX
    
    Dhrystone 2 using register variables        376783.7  1333463.8       35.4
    Double-Precision Whetstone                      83.1      634.1       76.3
    Execl Throughput                               188.3      238.4       12.7
    File Copy 1024 bufsize 2000 maxblocks         2672.0     4611.0       17.3
    File Copy 256 bufsize 500 maxblocks           1077.0     2275.0       21.1
    File Read 4096 bufsize 8000 maxblocks        15382.0    76111.0       49.5
    Pipe Throughput                             111814.6    72525.4        6.5
    Pipe-based Context Switching                 15448.6    19216.8       12.4
    Process Creation                               569.3      393.0        6.9
    Shell Scripts (8 concurrent)                    44.8       64.9       14.5
    System Call Overhead                        114433.5    37640.8        3.3
                                                                     =========
         FINAL SCORE                                                      15.9
    Jack.
      0 Not allowed!

  24. #24
    I'm still wondering how to interpret these benchmarks...
    Is the servint server really half the speed of the powervps in doing "real" tasks i..e mysql/web etc... or does the nature of the benchmark hit limits (or bursts) that the vps is set to control that real-world site tasks wouldnt hit. There are many optimizations that can be done to apache and mysql that I'm not sure that this benchmark can measure.

    Also I wish there was a benchmark that could be run on a shared host and a vps ... It would be interesting to see who would win that performance race.


    d
    Last edited by clearstr; 06-28-2005 at 12:12 AM.
      0 Not allowed!

  25. #25
    Power VPS 44.5 USD/m


    INDEX VALUES
    TEST BASELINE RESULT INDEX

    Dhrystone 2 using register variables 376783.7 2875638.6 76.3
    Double-Precision Whetstone 83.1 815.0 98.1
    Execl Throughput 188.3 649.8 34.5
    File Copy 1024 bufsize 2000 maxblocks 2672.0 5620.0 21.0
    File Copy 256 bufsize 500 maxblocks 1077.0 1689.0 15.7
    File Read 4096 bufsize 8000 maxblocks 15382.0 113337.0 73.7
    Pipe Throughput 111814.6 96013.7 8.6
    Pipe-based Context Switching 15448.6 42410.7 27.5
    Process Creation 569.3 612.0 10.8
    Shell Scripts (8 concurrent) 44.8 94.0 21.0
    System Call Overhead 114433.5 209930.4 18.3
    =========
    FINAL SCORE 27.2
      0 Not allowed!

  26. #26
    A2B2 30GB VPS Plan with cPanel Everything was running in the background IE...Apache Sql ect


    ==============================================================
    BYTE UNIX Benchmarks (Version 4.1-wht)
    System -- Linux kay.livemusiconline.net 2.4.20-021stab028.12.777-smp #1 SMP Fri May 13 16:18:38 MSD 2005 i686 athlon i386 GNU/Linux
    /dev/vzfs 30765625 1227103 29538522 4% /

    Start Benchmark Run: Tue Jun 28 16:12:15 BST 2005
    16:12:15 up 3 days, 23:06, 1 user, load average: 0.48, 0.48, 0.20

    End Benchmark Run: Tue Jun 28 16:23:31 BST 2005
    16:23:31 up 3 days, 23:17, 1 user, load average: 15.08, 6.29, 2.88


    INDEX VALUES
    TEST BASELINE RESULT INDEX

    Dhrystone 2 using register variables 376783.7 3004679.8 79.7
    Double-Precision Whetstone 83.1 741.8 89.3
    Execl Throughput 188.3 1353.5 71.9
    File Copy 1024 bufsize 2000 maxblocks 2672.0 15959.0 59.7
    File Copy 256 bufsize 500 maxblocks 1077.0 4481.0 41.6
    File Read 4096 bufsize 8000 maxblocks 15382.0 155361.0 101.0
    Pipe Throughput 111814.6 165461.9 14.8
    Pipe-based Context Switching 15448.6 54003.8 35.0
    Process Creation 569.3 4218.4 74.1
    Shell Scripts (8 concurrent) 44.8 312.8 69.8
    System Call Overhead 114433.5 298450.4 26.1
    =========
    FINAL SCORE 52.8
      0 Not allowed!

  27. #27
    Servint - $89 / month
    Red Hat Linux - 512MB / 2GB Burst Ram
    Bandwidth Included - 350GB

    ==============================================================
    BYTE UNIX Benchmarks (Version 4.1-wht)
    System -- Linux vps.spoofee.com 2.4.20-021stab028-smp-conntracks #1 SMP Fri Jun 24 13:41:19 MSD 2005 i686 unknown
    /dev/vzfs 20971520 4151958 16819562 20% /

    Start Benchmark Run: Wed Jun 29 14:47:09 PDT 2005
    2:47pm up 15:43, 1 user, load average: 1.25, 0.96, 0.91

    End Benchmark Run: Wed Jun 29 15:00:22 PDT 2005
    3:00pm up 15:56, 1 user, load average: 21.08, 9.82, 5.30


    INDEX VALUES
    TEST BASELINE RESULT INDEX

    Dhrystone 2 using register variables 376783.7 1785754.3 47.4
    Double-Precision Whetstone 83.1 499.7 60.1
    Execl Throughput 188.3 302.3 16.1
    File Copy 1024 bufsize 2000 maxblocks 2672.0 3733.0 14.0
    File Copy 256 bufsize 500 maxblocks 1077.0 1276.0 11.8
    File Read 4096 bufsize 8000 maxblocks 15382.0 65461.0 42.6
    Pipe Throughput 111814.6 51726.5 4.6
    Process Creation 569.3 426.9 7.5
    Shell Scripts (8 concurrent) 44.8 43.9 9.8
    System Call Overhead 114433.5 23665.6 2.1
    =========
    FINAL SCORE 13.7
    Spoofee
      0 Not allowed!

  28. #28
    Join Date
    Jul 2002
    Location
    Manchester, UK
    Posts
    774
    Running Xen kernel 2.6.10-xenU on my own VPS provisioning system (ProbonicVPS) :-

    Code:
    ==============================================================
    BYTE UNIX Benchmarks (Version 4.1-wht)
    System -- Linux config1.probonic.com 2.6.10-xenU #1 Thu Mar 10 19:37:47 GMT 2005 i686 i686 i386 GNU/Linux
    /dev/hda1              1758776   1210440    458992  73% /
    
    Start Benchmark Run: Wed Jun 29 16:02:54 EDT 2005
     16:02:54 up 5 days, 13:43,  1 user,  load average: 0.08, 0.02, 0.01
    
    End Benchmark Run: Wed Jun 29 16:14:56 EDT 2005
     16:14:56 up 5 days, 13:55,  1 user,  load average: 12.40, 5.54, 2.60
    
    
                         INDEX VALUES
    TEST                                        BASELINE     RESULT      INDEX
    
    Dhrystone 2 using register variables        376783.7  3567456.1       94.7
    Execl Throughput                               188.3     1530.7       81.3
    File Copy 1024 bufsize 2000 maxblocks         2672.0    50744.0      189.9
    File Copy 256 bufsize 500 maxblocks           1077.0    25115.0      233.2
    File Read 4096 bufsize 8000 maxblocks        15382.0   185589.0      120.7
    Pipe Throughput                             111814.6   604376.5       54.1
    Process Creation                               569.3     3952.0       69.4
    Shell Scripts (8 concurrent)                    44.8      158.9       35.5
    System Call Overhead                        114433.5   472087.7       41.3
                                                                     =========
         FINAL SCORE                                                      84.8
      0 Not allowed!

  29. #29
    Join Date
    Jul 2003
    Location
    Earth
    Posts
    1,692
    Just saw two or three results from ServInt users, and it's really amazing how low are the numbers... really bad performance, I'm afraid...

    It would be useful to know if they are VPS located at their new RAID-10 nodes though...
      0 Not allowed!

  30. #30
    Join Date
    Dec 2004
    Location
    New York, NY
    Posts
    10,574
    I don't have the exact benchmarks off hand but, I remember getting about ~ 55 on a RAID10 host node that was fairly empty. I then got around ~ 30 on a fairly filled host node.

    GSV
    MediaLayer, LLC - www.medialayer.com Learn how we can make your website load faster, translating to better conversion rates for your business!
    The pioneers of optimized web hosting, featuring LiteSpeed Web Server & SSD Storage - Celebrating 10 Years in Business
      0 Not allowed!

  31. #31
    Join Date
    Aug 2002
    Location
    Washington, D.C.
    Posts
    502
    Originally posted by Apolo
    Just saw two or three results from ServInt users, and it's really amazing how low are the numbers... really bad performance, I'm afraid...

    It would be useful to know if they are VPS located at their new RAID-10 nodes though...
    VPS CPU limits are dampened over time based on very high usage. This prevents one VPS from using all the system resources all day long. Because the UNIXBench test is primarily CPU and runs for 10-15 minutes, it gets dampened. Thus, test results aren't going to include high numbers. If we were to remove our dampening algorithms, we would look much cooler in this thread, but we'd be putting customers at risk.
    ServInt
    Providing Managed VPS for over 11 years
    Managed Dedicated Servers for almost 19 years!
    http://www.servint.net
      0 Not allowed!

  32. #32
    Join Date
    Jul 2003
    Location
    Earth
    Posts
    1,692
    Thanks for your explanation Reed. (hope you are )

    Would you mind to explain a bit more? 10-15 minutes does not seem too much time... or did I understand wrongly?

    As far as I know every VPS configuration allows certain amount of burstable resources, so from your statement I understand yours are certainly able to burst, but within a period of a couple of minutes only. (?)

    Thank you.
      0 Not allowed!

  33. #33
    Join Date
    Aug 2002
    Location
    Washington, D.C.
    Posts
    502
    Originally posted by Apolo
    Thanks for your explanation Reed. (hope you are )

    Would you mind to explain a bit more? 10-15 minutes does not seem too much time... or did I understand wrongly?

    As far as I know every VPS configuration allows certain amount of burstable resources, so from your statement I understand yours are certainly able to burst, but within a period of a couple of minutes only. (?)

    Thank you.
    I believe it begins dampening after about 5 minutes of steady, high activity. There are multiple levels of dampening, and based on usage and recent history certain CPU-intensive activities may trigger more than one level. Once the high activity relaxes, the CPU limits are raised back to normal. All this does is reduce burstable caps, it won't invate a VEs guaranteed resources.

    Even on a wide open dedicated server, UNIXBench typically takes 10-15 minutes. The point of our dampening is to ensure that runaway processes, abusive VEs, exploited applications, etc, don't impact all VEs on a host machine. Our intention isn't to dampen UNIXBench itself, that's just an unfortunate side effect.
    ServInt
    Providing Managed VPS for over 11 years
    Managed Dedicated Servers for almost 19 years!
    http://www.servint.net
      0 Not allowed!

  34. #34
    Join Date
    Jun 2005
    Location
    The Server Rooms :)
    Posts
    163
    This is on one of my dev boxes to test game servers, etc..

    ==============================================================
    BYTE UNIX Benchmarks (Version 4.1-wht)
    System -- Linux vps2.turbulentmedia.com 2.4.20-021stab028.3.777-smp #1 SMP Wed Feb 2 21:32:05 MSK 2005 i686 athlon i386 GNU/Linux
    /dev/vzfs 15728640 2229177 13499463 15% /

    Start Benchmark Run: Thu Jun 30 13:26:21 CDT 2005
    13:26:21 up 2 days, 22:52, 1 user, load average: 0.31, 0.10, 0.03

    End Benchmark Run: Thu Jun 30 13:39:56 CDT 2005
    13:39:56 up 2 days, 23:06, 1 user, load average: 12.45, 5.44, 2.61


    INDEX VALUES
    TEST BASELINE RESULT INDEX

    Dhrystone 2 using register variables 376783.7 3818827.2 101.4
    Double-Precision Whetstone 83.1 805.8 97.0
    Execl Throughput 188.3 1858.9 98.7
    File Copy 1024 bufsize 2000 maxblocks 2672.0 27008.0 101.1
    File Copy 256 bufsize 500 maxblocks 1077.0 8224.0 76.4
    File Read 4096 bufsize 8000 maxblocks 15382.0 177631.0 115.5
    Pipe Throughput 111814.6 242429.4 21.7
    Pipe-based Context Switching 15448.6 74950.2 48.5
    Process Creation 569.3 3809.2 66.9
    Shell Scripts (8 concurrent) 44.8 357.9 79.9
    System Call Overhead 114433.5 460820.3 40.3
    =========
    FINAL SCORE 69.9
      0 Not allowed!

  35. #35
    VPS2 w/ LiquidWeb

    Start Benchmark Run: Fri Jul 1 16:00:32 MST 2005
    16:00:32 up 10 days, 13:09, 1 user, load average: 0.33, 0.14, 0.04

    End Benchmark Run: Fri Jul 1 16:12:13 MST 2005
    16:12:13 up 10 days, 13:20, 1 user, load average: 15.89, 6.67, 3.01


    INDEX VALUES
    TEST BASELINE RESULT INDEX

    Dhrystone 2 using register variables 376783.7 4698080.6 124.7
    Double-Precision Whetstone 83.1 761.0 91.6
    Execl Throughput 188.3 1613.7 85.7
    File Copy 1024 bufsize 2000 maxblocks 2672.0 23751.0 88.9
    File Copy 256 bufsize 500 maxblocks 1077.0 4991.0 46.3
    File Read 4096 bufsize 8000 maxblocks 15382.0 184453.0 119.9
    Pipe Throughput 111814.6 287711.9 25.7
    Pipe-based Context Switching 15448.6 66598.4 43.1
    Process Creation 569.3 4024.2 70.7
    Shell Scripts (8 concurrent) 44.8 353.4 78.9
    System Call Overhead 114433.5 354756.3 31.0
    =========
    FINAL SCORE 65.3
      0 Not allowed!

  36. #36
    Join Date
    Dec 2004
    Location
    New York, NY
    Posts
    10,574
    EuroVPS: VPS 200 w/Plesk
    Code:
    ==============================================================
    BYTE UNIX Benchmarks (Version 4.1-wht)
    System -- Linux nl.medialayer.com 2.4.20-021stab028.12.777-enterprise #1 SMP Fri May 13 15:50:02 MSD 2005 i686 i686 i386 GNU/Linux
    /dev/vzfs             15539895    129138  15410757   1% /
    
    Start Benchmark Run: Mon Jul  4 22:50:45 EEST 2005
     22:50:46  up 42 min,  1 user,  load average: 0.09, 0.08, 0.11
    
    End Benchmark Run: Mon Jul  4 23:01:11 EEST 2005
     23:01:11  up 52 min,  2 users,  load average: 13.67, 5.72, 2.56
    
    
                         INDEX VALUES
    TEST                                        BASELINE     RESULT      INDEX
    
    Dhrystone 2 using register variables        376783.7  5904549.0      156.7
    Double-Precision Whetstone                      83.1      532.7       64.1
    Execl Throughput                               188.3     2371.5      125.9
    File Copy 1024 bufsize 2000 maxblocks         2672.0    12873.0       48.2
    File Copy 256 bufsize 500 maxblocks           1077.0     5766.0       53.5
    File Read 4096 bufsize 8000 maxblocks        15382.0   289147.0      188.0
    Pipe Throughput                             111814.6   279523.4       25.0
    Pipe-based Context Switching                 15448.6    81958.9       53.1
    Process Creation                               569.3     2667.2       46.9
    Shell Scripts (8 concurrent)                    44.8      463.7      103.5
    System Call Overhead                        114433.5   133089.6       11.6
                                                                     =========
         FINAL SCORE                                                      61.3
    Very nice, if I say so myself.

    -GSV
    MediaLayer, LLC - www.medialayer.com Learn how we can make your website load faster, translating to better conversion rates for your business!
    The pioneers of optimized web hosting, featuring LiteSpeed Web Server & SSD Storage - Celebrating 10 Years in Business
      0 Not allowed!

  37. #37
    Originally posted by BobFarmer
    There are multiple levels of dampening, and based on usage and recent history certain CPU-intensive activities may trigger more than one level.
    Does this mean that the dampening occurs on a per-VPS basis or is the profile Servint wide?

    Originally posted by BobFarmer
    The point of our dampening is to ensure that runaway processes, abusive VEs, exploited applications, etc, don't impact all VEs on a host machine. Our intention isn't to dampen UNIXBench itself, that's just an unfortunate side effect.
    If I want to run a CPU intensive script that takes a 7 minute period to execute, then it would apparently also suffer from unfortunate side-affects, eh?

    Originally posted by BobFarmer
    All this does is reduce burstable caps, it won't invate a VEs guaranteed resources.
    With no idea how many clients are on a box or guaranteed CPU minimums, it's really hard to ascertain each VPS's guaranteed resources. It's unpublished data, and from the answers I have seen, busy machines have less clients and slower servers may have more clients. So guaranteed resources varies from machine to machine, client to client regardless of the package selected.

    Am I off base?
    Last edited by autodelete; 07-04-2005 at 08:09 PM.
      0 Not allowed!

  38. #38
    Join Date
    Jun 2005
    Location
    European Union
    Posts
    387
    As far as I know every VPS configuration allows certain amount of burstable resources, so from your statement I understand yours are certainly able to burst, but within a period of a couple of minutes only. (?)
    I am curious, how is this being accomplished in Virtuozzo, as the limits available to control CPU usage are "CPUUNITS" and "CPULIMIT".

    From the training we received from SW-Soft, we were directed not to place limits percentage-wise on VE's on a server with more than 5-6 VE's, as the only functionality of the CPULIMIT is to keep customers that end up on empty machines from believing that they are having "free reign" of an 8 processor machine.

    Now, as far as dampening, and stopping bursting, this doesn't make sense - it sounds like you've got static % CPULIMIT parameters set, in which case the performance will *always* be the same.

    I am actually curious about this from an engineering standpoint, just because this affects our organization as well; it does happen that we will get a rogue user. Last week we had a guy that downloaded a program called "Stress" from one of the Linux Benchmarking centers, and executed it on a VPS w/o a time limit, this spawned hundreds of processes which drained the entire VZ node until we shut down his server.

    The trick here is to have constant monitoring to know when this happens, but beyond this, how are you "dampening" after 5-6 minutes? I want to know what the parameter in VZMC/VZ is that can do this after a "set time", this is extremely helpful for us.

    I have looked in the manuals/references that SW-Soft provides to us and can find no such feature which could limit bursting after a set time.


    Thanks in advance to Servint for sharing the secret,
    Igor
    EuroVPS - Europe's Fully Managed Hosting
    Established 2004 - The European Hosting Authority
      0 Not allowed!

  39. #39
    You all lose:

    Dhrystone 2 using register variables 376783.7 5137271.7 136.3
    Double-Precision Whetstone 83.1 1275.6 153.5
    Execl Throughput 188.3 2178.4 115.7
    File Copy 1024 bufsize 2000 maxblocks 2672.0 24995.0 93.5
    File Copy 256 bufsize 500 maxblocks 1077.0 8499.0 78.9
    File Read 4096 bufsize 8000 maxblocks 15382.0 248827.0 161.8
    Pipe Throughput 111814.6 384982.6 34.4
    Pipe-based Context Switching 15448.6 82568.8 53.4
    Process Creation 569.3 5985.7 105.1
    Shell Scripts (8 concurrent) 44.8 490.1 109.4
    System Call Overhead 114433.5 363764.1 31.8
    =========
    FINAL SCORE 86.2
      0 Not allowed!

  40. #40
    Join Date
    Jul 2005
    Location
    Grand Rapids, MI
    Posts
    2
    Results of VPSCP1 Plan at Vpscolo.com:

    BYTE UNIX Benchmarks (Version 4.1-wht)
    System -- Linux host.powerkeyspub.com 2.4.23-secure-1um #4 Thu Mar 4 18:53:28 EST 2004 i686 i686 i386 GNU/Linux
    /dev/ubd/0 8256952 3581640 4339768 46% /

    Start Benchmark Run: Tue Jul 5 11:40:36 EDT 2005
    11:40:36 up 21 days, 17:42, 2 users, load average: 0.80, 0.76, 0.82

    End Benchmark Run: Tue Jul 5 11:53:58 EDT 2005
    11:53:58 up 21 days, 17:55, 2 users, load average: 23.62, 13.35, 7.25


    INDEX VALUES
    TEST BASELINE RESULT INDEX

    Dhrystone 2 using register variables 376783.7 2560220.2 67.9
    Double-Precision Whetstone 83.1 565.1 68.0
    Execl Throughput 188.3 130.2 6.9
    File Copy 1024 bufsize 2000 maxblocks 2672.0 2491.0 9.3
    File Copy 256 bufsize 500 maxblocks 1077.0 957.0 8.9
    File Read 4096 bufsize 8000 maxblocks 15382.0 39897.0 25.9
    Pipe Throughput 111814.6 20407.3 1.8
    Process Creation 569.3 103.3 1.8
    Shell Scripts (8 concurrent) 44.8 14.5 3.2
    System Call Overhead 114433.5 9777.8 0.9
    =========
    FINAL SCORE 7.6

    I've been complaining for weeks about slow performance, and now I have numbers to back it up.

    - Alan Tutt
      0 Not allowed!

Page 1 of 21 1234 11 ... LastLast

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •