SPONSORS:






User Tag List

Thanks Thanks:  0
Likes Likes:  0
Dislikes Dislikes:  0
Page 1 of 2 12 LastLast
Results 1 to 10 of 16
  1. #1
    Senior Member
    Join Date
    May 2001
    Posts
    265
    Post Thanks / Like
    Mentioned
    0 Post(s)
    Tagged
    0 Thread(s)
    Total Downloaded
    0

    Max number of user

    We have a licence for 100 user and client needs data for 300-600 users.
    I was thinking if i take out the thinktime that would bumpup the traffic (throughput /sec)

    Please correct me if I am wrong making this assumption:

    1. Do the loadtest using 100 users (with thinktime) and get the throughput.
    2. Take the throughput and multiply it with 3 to get the data for 300 users.
    3. Finally, do the same test of 100 users (without thinktime) and see if I can match that throughput rate that I got in step 2.
    If it matches does it mean that I am able to generate traffic for 300 users?

    Just curious to know,how does how these large companies i.e. cnn.com does the load test for 100,000 users do they extrapolate their results.

    [ 05-25-2005, 12:14 PM: Message edited by: KS ]

  2. #2
    Advanced Member LauraScharp's Avatar
    Join Date
    Aug 2002
    Location
    Huntington Beach, Ca. USA
    Posts
    725
    Post Thanks / Like
    Mentioned
    0 Post(s)
    Tagged
    0 Thread(s)
    Total Downloaded
    0

    Re: Max number of user

    Hi KS,

    I've seen this question come up from time to time - here is a thread that addresses it very well...

    http://www.qaforums.com/cgi-bin/foru...=003673#000006

    In regards to how the larger companies do this - I think that in many cases, they do run full load tests. There are companies (Mercury, for one, IBM for another) that will allow you to purchase short term licensing and rent time on their load generators to accommodate this kind of testing. It is not inexpensive, but to quote a popular commercial... "Knowing that your site can handle 100,000 concurrent users? Priceless!" [img]images/icons/smile.gif[/img]
    Laura Scharp
    SoCalGal - Defender of end user response times!

  3. #3
    Senior Member
    Join Date
    Aug 2003
    Location
    London
    Posts
    190
    Post Thanks / Like
    Mentioned
    0 Post(s)
    Tagged
    0 Thread(s)
    Total Downloaded
    0

    Re: Max number of user

    Why not factor down the thinktime instead of removing it altogether?

    So if total thinktime is estimated as 60 seconds, then use 20 seconds of think time to emulate 300 users in your next test.

    The other problem you will face is testing the number of concurrent connections.

  4. #4
    Moderator JakeBrake's Avatar
    Join Date
    Dec 2000
    Location
    St. Louis - Year 2025
    Posts
    15,609
    Post Thanks / Like
    Mentioned
    0 Post(s)
    Tagged
    0 Thread(s)
    Total Downloaded
    0

    Re: Max number of user

    Originally posted by mongoose:
    Why not factor down the thinktime instead of removing it altogether?

    ...
    <font size="2" face="Verdana, Arial, Helvetica">For the reasons listed here:

    http://www.qaforums.com/cgi-bin/foru...=003673#000006

  5. #5
    Senior Member
    Join Date
    May 2001
    Posts
    265
    Post Thanks / Like
    Mentioned
    0 Post(s)
    Tagged
    0 Thread(s)
    Total Downloaded
    0

    Re: Max number of user

    Thank you for the reply.
    I just wanted to get your ideas to figure out the correlation of thinktime and no of users.
    I read the other threads and what I understand is that factoring down thinktime may increase the the # of users but you can't be sure by how many.

    If you run a test with 1 VUser then change it 100 VUser.
    As far as the load, only thing you have increased here is it throughput/sec. Is that correct?

  6. #6
    Advanced Member LauraScharp's Avatar
    Join Date
    Aug 2002
    Location
    Huntington Beach, Ca. USA
    Posts
    725
    Post Thanks / Like
    Mentioned
    0 Post(s)
    Tagged
    0 Thread(s)
    Total Downloaded
    0

    Re: Max number of user

    Originally posted by KS:
    If you run a test with 1 VUser then change it 100 VUser.
    As far as the load, only thing you have increased here is it throughput/sec. Is that correct?
    <font size="2" face="Verdana, Arial, Helvetica">Not in my opinion. From a pure number of transactions per second, maybe, but 100 users will utilize more system resources than 1 user. Even 100 users doing nothing will require more resources than 1 user doing nothing. You have more sessions, more attachments to the database, more of an opportunity for locking in the database... the list goes on.

    One question you absolutely cannot answer in the above scenario is this - "Can 100 users login to the system". The only way to say for sure that 100 users can log in is to have 100 users log in.
    Laura Scharp
    SoCalGal - Defender of end user response times!

  7. #7
    Member
    Join Date
    Jan 2002
    Location
    Michigan, USA
    Posts
    44
    Post Thanks / Like
    Mentioned
    0 Post(s)
    Tagged
    0 Thread(s)
    Total Downloaded
    0

    Re: Max number of user

    Removing think time can kill an application. A system will perform better with more users than it will with less think time.

    Example:
    All transactions take 5 seconds.
    Base: 50 VU w/ 5s Think = 5tps
    Double to: (In theory)
    50 VU w/ 0s Think = 10tps
    100 VU w/ 5s Think = 10tps

    Especially on a complex, data-driven app, the 100 VU will outperform the 50 VU _every_ time. The reason is that the number of connections/session is going through the roof. For a single session, everything happens within the same thread/set of threads. Because that single thread is constantly changing back and forth between doing the real work and opening/closing connections, it causes a higher demand on your CPUs, and you end up flooding the system.

    Though using more VU still opens/closes the same number of connections, the server will use more threads to do it (Because of more simultaneous sessions), and each individual thread isn't switching back and forth as often.

    Think of it this way:
    100 people will get more done working 40 hours a week than 50 people will working 80 hours a week.
    Jay Harris
    Cpt. LoadTest
    http://www.cptloadtest.com

  8. #8
    Member
    Join Date
    Jul 2002
    Posts
    48
    Post Thanks / Like
    Mentioned
    0 Post(s)
    Tagged
    0 Thread(s)
    Total Downloaded
    0

    Re: Max number of user

    Originally posted by Cpt. LoadTest:
    Removing think time can kill an application. A system will perform better with more users than it will with less think time.

    Example:
    All transactions take 5 seconds.
    Base: 50 VU w/ 5s Think = 5tps
    Double to: (In theory)
    50 VU w/ 0s Think = 10tps
    100 VU w/ 5s Think = 10tps

    <font size="2" face="Verdana, Arial, Helvetica">good, very good!!
    I once calculated it like this.

  9. #9
    Moderator
    Join Date
    Sep 2001
    Location
    Boston, MA
    Posts
    4,348
    Post Thanks / Like
    Mentioned
    0 Post(s)
    Tagged
    0 Thread(s)
    Total Downloaded
    0

    Re: Max number of user

    Originally posted by Cpt. LoadTest:
    Removing think time can kill an application. A system will perform better with more users than it will with less think time.

    Especially on a complex, data-driven app, the 100 VU will outperform the 50 VU _every_ time. The reason is that the number of connections/session is going through the roof. For a single session, everything happens within the same thread/set of threads. Because that single thread is constantly changing back and forth between doing the real work and opening/closing connections, it causes a higher demand on your CPUs, and you end up flooding the system.

    Though using more VU still opens/closes the same number of connections, the server will use more threads to do it (Because of more simultaneous sessions), and each individual thread isn't switching back and forth as often.
    <font size="2" face="Verdana, Arial, Helvetica">I don't fully agree with this. I think you are citing a specific example and I would be hesitant to look at this as a general principle for characterizing load/capacity on all systems (or "every" time as you stated). A web application or service may use a complex set of queues and thread pools that would make it impossible to come to this conclusion without knowing more about the architecture.
    Corey Goldberg
    Homepage: goldb.org
    Twitter: twitter.com/cgoldberg
    Google+: gplus.to/cgoldberg

  10. #10
    Member
    Join Date
    Jan 2002
    Location
    Michigan, USA
    Posts
    44
    Post Thanks / Like
    Mentioned
    0 Post(s)
    Tagged
    0 Thread(s)
    Total Downloaded
    0

    Re: Max number of user

    Corey,

    Yes, perhaps it was an over-generalization, but it has been my experience to date. Granted, my numbers (50VU vs 100VU) were simplified to make it easy to explain, but this has been my experience throughout my career, including everything from testing a single server running IIS and a local database, to a complex, 4-tier, farmed environment with an AS400 at one end.

    My only qualifier on my scenario: The system must be able to handle 100 simultaneous connections in the first place.
    Jay Harris
    Cpt. LoadTest
    http://www.cptloadtest.com

 

 
Page 1 of 2 12 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
  •  
Search Engine Optimisation provided by DragonByte SEO v2.0.36 (Pro) - vBulletin Mods & Addons Copyright © 2016 DragonByte Technologies Ltd.
Resources saved on this page: MySQL 9.38%
vBulletin Optimisation provided by vB Optimise v2.6.4 (Pro) - vBulletin Mods & Addons Copyright © 2016 DragonByte Technologies Ltd.
User Alert System provided by Advanced User Tagging v3.2.8 (Pro) - vBulletin Mods & Addons Copyright © 2016 DragonByte Technologies Ltd.
vBNominate (Lite) - vBulletin Mods & Addons Copyright © 2016 DragonByte Technologies Ltd.
Feedback Buttons provided by Advanced Post Thanks / Like (Pro) - vBulletin Mods & Addons Copyright © 2016 DragonByte Technologies Ltd.
Username Changing provided by Username Change (Free) - vBulletin Mods & Addons Copyright © 2016 DragonByte Technologies Ltd.
BetaSoft Inc.
Digital Point modules: Sphinx-based search
All times are GMT -8. The time now is 06:08 PM.

Copyright BetaSoft Inc.