locked
understanding the "Test results: Throughput by farm configuration" table RRS feed

  • Question

  • Hi there

    Can someone tell me how to read this table?

    The table below is from: http://technet.microsoft.com/en-us/library/cc261795(office.12).aspx

    Farm configuration RPS   Total number of user connections              

     

     

     

    Light usage

     

    Typical usage

     

    Heavy usage

     

    Extreme usage

     

     

    Mix

    Read

    Mix

    Read

    Mix

    Read

    Mix

    Read

    Mix

    Read

    1 by 1

    50

    100

    90,000

    180,000

    50,000

    100,000

    30,000

    60,000

    15,000

    As I understand it this means that for the given hardware used in testing and for a 1x1 configuration - ... well what? I don't get it

    RPS means requests per second

    The number of user connections is a fixed parameter that was used during testing

    Does it mean that for 90,000 user connections doing mixed operations the config supports 50 RPS? And for 180,000 connections doing Read operations it supports 100 RPS? So where does the light usage , medium usage etc come in to it?

    Can anyone enlighten me?

    Cheers

    Jonj


    • Edited by jonjames Thursday, June 9, 2011 10:54 AM more detail
    Thursday, June 9, 2011 10:49 AM

Answers

  • Hi joinjames,

     

    I have read the article seriously, and I’m also very interested in this issue.

     

    The first “1” of “1 by 1” represents 1 Web server computer, and the second “1” represents 1(clustered or mirrored) database server computer. You can get more detailed information from the part of “Hardware recommendations” in the article.

     

    And according to my understanding, RPS is the prerequisite, it is depends on your Farm configuration and it is for all of the four situation(light, typical, heavy, extreme usage). With that premise, under different user load request rate, the test results of the Read number and Total number of user connections will reach different number.

     

    Thanks & Regards,

    Peng Lei

    • Marked as answer by David HM Friday, June 17, 2011 2:01 AM
    Friday, June 10, 2011 3:49 AM
  • Hi jonjames,

     

    According to my understanding, these are the average performance of the system under different configuration. These do not represent the maximum or minimum values, just some test data. Perhaps in the next test these data will be slightly larger or smaller.

     

    The use of these data is to help you choose a suitable system for your condition. With the least money, get the most performance. In that article, you can see a graph which shows changes in throughput for both read-write and read-only operations when the number of front-end Web servers changes. You can see that the performance increases quickly before the number of Front End Web Server is 3, so you can paid less money to get more performance. But after that, performance improvement obtained no more than before.

     

    Thanks & Regards,

    Peng Lei

    • Marked as answer by David HM Friday, June 17, 2011 2:01 AM
    Friday, June 10, 2011 8:16 AM

All replies

  • Hi joinjames,

     

    I have read the article seriously, and I’m also very interested in this issue.

     

    The first “1” of “1 by 1” represents 1 Web server computer, and the second “1” represents 1(clustered or mirrored) database server computer. You can get more detailed information from the part of “Hardware recommendations” in the article.

     

    And according to my understanding, RPS is the prerequisite, it is depends on your Farm configuration and it is for all of the four situation(light, typical, heavy, extreme usage). With that premise, under different user load request rate, the test results of the Read number and Total number of user connections will reach different number.

     

    Thanks & Regards,

    Peng Lei

    • Marked as answer by David HM Friday, June 17, 2011 2:01 AM
    Friday, June 10, 2011 3:49 AM
  • Hi Peng Lei

    Thanks very much for responding - can you tell me then if this is correct:

    So for a 1x1 configuration serving 50 requests per second under light usage this should support 90,000 user connections doing Mixed Read/Write operations or 180,000 User Connections doing Read operations

    Is that right?

    Cheers

    Jonj

    Friday, June 10, 2011 7:24 AM
  • Hi jonjames,

     

    According to my understanding, these are the average performance of the system under different configuration. These do not represent the maximum or minimum values, just some test data. Perhaps in the next test these data will be slightly larger or smaller.

     

    The use of these data is to help you choose a suitable system for your condition. With the least money, get the most performance. In that article, you can see a graph which shows changes in throughput for both read-write and read-only operations when the number of front-end Web servers changes. You can see that the performance increases quickly before the number of Front End Web Server is 3, so you can paid less money to get more performance. But after that, performance improvement obtained no more than before.

     

    Thanks & Regards,

    Peng Lei

    • Marked as answer by David HM Friday, June 17, 2011 2:01 AM
    Friday, June 10, 2011 8:16 AM
  • Thanks Peng Lei - I do understand that this is just test data and each environment will vary in its' requirements.

    The reason I'm working to understand these tables is because I need to work out a model/formula that I can use for capacity planning our future SharePoint deployments

    There used to be a tool for this: SPCAP - but it has been discontinued and except for these TechNet articles I don't know what else is now being used to plan this aspect of a SharePoint deployment

    I suspect most companies are not going to have the time/money to experiment with different topologies under different loads generated by expensive load generating tools - the company I work for certainly won't. They want a formula they can pump numbers (Total users, concurrency levels, request rates, acceptable response times) in one end and get Farm configs out the other

    Thanks for all your help - I think I'm reading the table correctly now - next step is turn it into a formula -

    Cheers

    Jonj

    Friday, June 10, 2011 9:46 AM