Threads and Sequence of execution

[Christiaan Lamprecht <ch...@googlemail.com>]

Hi all,

Just a simple-ish "sequence of execution" question if anyone can help;

Test Plan
  Thread Group (users 5, ramp 3, count 500)
     HTTP Request HTTPClient (SSL request - KeepAlive enabled)
Uniform Random Timer (Offset 100, dev 5)


>From the above nested test plan (and documentation) I assume:
* 5 threads are created in the space of 3 seconds
* The Thread Group contents are executed concurrently 500 times for
each thread but only 5 threads are ever created. i.e. "loop count"
applies only to the Thread Group elements
* Each of the 500 requests are separated by 100ms
* Only 5 SSL sessions are established with the server (unless
time-outs etc. interfere)


If the above is correct can I test the server for a scenario where for
example; the server gets 10 new users (i.e. 10 threads or 10 SSL
connections) each second and each user makes a 100 requests in that
second and repeat the experiment for 3 mins to get a good average
response time?

Perhaps something like...

Test Plan
  Thread Group (users 100, ramp 10, count 100)
     HTTP Request HTTPClient (SSL request - KeepAlive enabled)
Uniform Random Timer (Offset 10, dev 2)

... but I need an average, over 3 mins, for each set of 10 users/sec
(which need to be in separate files so that I can calculate the
averages)


Though perhaps I'm going against the grain as it seems that "ramp-up
time" is there to ease the burden of creating many new threads rather
than using it to create many new threads!?


Many thanks
Christiaan

---------------------------------------------------------------------
To unsubscribe, e-mail: jmeter-user-unsubscribe@jakarta.apache.org
For additional commands, e-mail: jmeter-user-help@jakarta.apache.org

Re: Threads and Sequence of execution

[Christiaan Lamprecht <ch...@googlemail.com>]

Please ignore the second half of my previous post (also included at
the end of this post). The solution is below. Also, the use of
"ramp-up" seems to suggest that creating threads are quite expensive,
so still not sure if the solution below is the right approach to
control the number of new SSL sessions.

Solution is;

Test Plan
 Thread Group (users 1800, ramp 180, count 100)
    HTTP Request HTTPClient (SSL request - KeepAlive enabled)
    Simple Data Writer
 Thread Group (users 1800, ramp 90, count 100)
    HTTP Request HTTPClient (SSL request - KeepAlive enabled)
    Simple Data Writer
 .......
 .......
 .......
Uniform Random Timer (Offset 10, dev 2)



Thanks, next time I'll post when I'm not sleep deprived!
Christiaan


> If the above is correct can I test the server for a scenario where for
> example; the server gets 10 new users (i.e. 10 threads or 10 SSL
> connections) each second and each user makes a 100 requests in that
> second and repeat the experiment for 3 mins to get a good average
> response time?
>
> Perhaps something like...
>
> Test Plan
>   Thread Group (users 100, ramp 10, count 100)
>      HTTP Request HTTPClient (SSL request - KeepAlive enabled)
> Uniform Random Timer (Offset 10, dev 2)
>
> ... but I need an average, over 3 mins, for each set of 10 users/sec
> (which need to be in separate files so that I can calculate the
> averages)
>
>
> Though perhaps I'm going against the grain as it seems that "ramp-up
> time" is there to ease the burden of creating many new threads rather
> than using it to create many new threads!?
>
>
> Many thanks
> Christiaan
>

---------------------------------------------------------------------
To unsubscribe, e-mail: jmeter-user-unsubscribe@jakarta.apache.org
For additional commands, e-mail: jmeter-user-help@jakarta.apache.org