Results 1 to 12 of 12
  1. #1
    Join Date
    Apr 2004
    Location
    Australia
    Posts
    448

    * 1000.00 Load for Dual Xeon? Normal?

    Now I have a client who claims his server runs at 1000.00 and thats considered almost normal, now he says I should change my script from the 1.0 = 1CPU rule.

    Which means for a server like his I would need to change it to 250.00 Load = 1CPU..
    And this would be crazy as I know Apache would crash before the load even reached 50% with the calculations set like that.

    Whats everyone else's view?
    What do you consider a good load value for each CPU?

    My Current Values: 1CPU = 1.0, 2CPU's = 2.0, Dual Xeon = 4.0

  2. #2
    Join Date
    Jul 2003
    Location
    Goleta, CA
    Posts
    5,550
    Well FreeBSD can operate with high load averages but 1000 ? I would think the server would crash long before it got that high.

    *checks site5 to make sure
    Patron: I'd like my free lunch please.
    Cafe Manager: Free lunch? Did you read the fine print stating it was an April Fool's joke.
    Patron: I read the same way I listen, I ignore the parts I don't agree with. I'm suing you for false advertising.
    Cafe Owner: Is our lawyer still working pro bono?

  3. #3
    Join Date
    Dec 2001
    Location
    127.0.0.1
    Posts
    3,640
    Your estimates sounds about right.
    Simpli Networks, LLC :: http://www.simplinetworks.com :: Proudly 100% Owned.
    Providing Affordable Managed Cloud/VPS Servers & Server Management Solutions.
    We offer REAL 24x7x365 in-house support - proudly serving our customers since 2005!
    Want to learn more? Give us a call - +1 (844) 4SIMPLI or email sales[@]simplinetworks.com today!

  4. #4
    Server load at 1000 ? I don't think i can even access the server. A load of 100 will get the whole thing crawl like a turtle(maybe even slower )
    Affordable Managed Hosting Solutions for Professional & Business since 2001
    Mxhub.com - Global : USA - UK - Canada - Europe - Asia Pacific


  5. #5
    Join Date
    Apr 2004
    Location
    Australia
    Posts
    448
    Just a quick quote here...

    my server can hit load 30 and still be running fine so this seems incorrect, also i have mail servers that can hit a load of 500-1500 and still be running fine
    Really when I read that I didn't believe it. He says he cant believe I develop the script I do, if I don't even know this! even though I have read countless articles on server load and ended up deciding that 1.0 was around the average for 1CPU.

  6. #6
    That's amazing. Is he sitting beside the server? ^_^
    Affordable Managed Hosting Solutions for Professional & Business since 2001
    Mxhub.com - Global : USA - UK - Canada - Europe - Asia Pacific


  7. #7
    Join Date
    Apr 2004
    Location
    Australia
    Posts
    448
    Maybe its a government super computer 0.o

    I was mainly just trying to find out if I was right or not, I mean wouldn't want all my clients thinking there servers were overloaded if they were not...

  8. #8
    Join Date
    Nov 2003
    Location
    Newport Beach, CA
    Posts
    2,920
    he's definitely calculating things differently. The way you wrote your script is the correct way to establish percentage of use (i bought your script).

    As can be seen here:
    http://www.e-referrer.com/sysinfo/
    Show your reciprocal links on your website. eReferrer

  9. #9
    Join Date
    May 2006
    Location
    /home/India/Kolkata
    Posts
    306
    for a dual zeon with ht, 10.00 should be the cutoff load

  10. #10
    Join Date
    Jan 2004
    Posts
    654
    Quote Originally Posted by scribby
    Now I have a client who claims his server runs at 1000.00 and thats considered almost normal, now he says I should change my script from the 1.0 = 1CPU rule.

    Which means for a server like his I would need to change it to 250.00 Load = 1CPU..
    And this would be crazy as I know Apache would crash before the load even reached 50% with the calculations set like that.

    Whats everyone else's view?
    What do you consider a good load value for each CPU?

    My Current Values: 1CPU = 1.0, 2CPU's = 2.0, Dual Xeon = 4.0
    i dont know so i wont comment on your question but id like to say your script is beautifull and im buying it!

    brilliant

  11. #11
    Join Date
    Apr 2004
    Location
    Australia
    Posts
    448
    well personally i find it embarassing to have to explain to my server administrators that the fact that the script shows such a low load is that the coder of the script doesnt believe load can go above 2.0, they are laughing at this.

    hostgator dont keep their servers at 10, they keep them at 3-5, however they dont investigate a server for moderate load until it reaches 10

    the majority of server administrators are not in the webhosting community, they are in administration/corp

    and what do you mean about it being abused, i bought the script, i should be able to edit this value if i want (apart from the fact that you encrypt the script for no reason)
    And people wonder why I'm going to be introducing paid support...

    Completely changes his view, now all of a sudden he saying around 3-5 when before he was giving me values in the thousands.

    I said the value cannot be changed due to the fact that it would be abused by hosts trying to pass there servers off as having load loads while in fact they have large loads.

    I knew from the start a server can go over 2.0, no need to be insulting...

    I think most people would agree with me, if you have 1 CPU and you get to 1.0 Load constantly that may be alright (although I still wouldn't run it like that) but once you start going over that 1.0 load then your server will have slower response times.

  12. #12
    Join Date
    Jan 2003
    Location
    Lake Arrowhead, CA
    Posts
    789
    There is really no way to make the numbers comparable between any two different systems. In the simplest sense, load averages indicate how many processes are waiting on CPU time. If you had eight way dual-core SMP and large numbers of very fast processes, then you could have extremely high load averages without trouble whereas a single CPU running very large, slow processes however might be completely shut down at a tiny fraction of that load.

    1.0 per physical CPU will tell you if the CPU is waiting, but it won't tell you whether that waiting has any effect whatsoever on content delivery.

    If you want to be a bit more accurate, you could run a benchmark process during script install which measures how long it takes the current system to execute specific (N*CPU) high load tasks and use that to modify your numbers, but even then there are so many factors (memory, disk I/O, network, etc.) that the results would still be far from reliable.
    http://www.srohosting.com
    Stability, redundancy and peace of mind

Posting Permissions

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