Well put.

The thread count, ramp-up and loop counts can be defined using properties - e.g.

${__P(threads)}
${__P(rampup)}
...

jmeter -Jthreads=10 -Jrampup=5 etc

S.
On 11/04/07, Matthew Coventon <[EMAIL PROTECTED]> wrote:
Sorry I sent an incomplete email by accident.  Let me reiterate:

My approach to this is to not automate it in JMeter.  Generally you don't
need such a fine granularity.  For example if you estimate that your system
can handle up to 100 concurrent users you don't need to know how a page
performs for 1, 2, 3, 4, 5, ... 97, 98, 99, 100.  Instead test with 1, 10,
25, 50, 75, 100, and 125 concurrent users.

In general for a target load X, do:
- Single user.
- Benchmark at 10 or 15% of X
- Load test at 25% of X
- Load test at 50% of X
- Load test at 75% of X
- Load test at 100% of X
- Load test at 125% of X

That's only seven tests and is sufficient information to create a
degradation curve.  For more info check out this document:
http://www.perftestplus.com/resources/UENM10.pdf

Regards,

Matt C.

-----Original Message-----
From: Matthew Coventon [mailto:[EMAIL PROTECTED]
Sent: Wednesday, April 11, 2007 9:01 AM
To: 'JMeter Users List'
Subject: RE: cuncurrency - responsetime chart

My approach to this is to not automate it in JMeter.  Generally you don't
need such a fine granularity.  For example if you estimate that your system
can handle up to 100 concurrent users you don't need to know how a page
performs for 1, 2, 3, 4, 5, ... 97, 98, 99, 100.  Instead test with 1, 10,
25, 50, 75, 100, and 125 concurrent users.

In general for a target load X test:
- Single user.
- Benchmark at 10 or 15% of X
- Load test at 25% of X
- Load test at 25% of X
- Load test at 25% of X
-----Original Message-----
From: SunWuKung [mailto:[EMAIL PROTECTED]
Sent: Wednesday, April 11, 2007 8:43 AM
To: [email protected]
Subject: Re: cuncurrency - responsetime chart


Thanks for this, but ...
As you say the ramp-up gradually increase the load, so when I look up the
response time for a page I can only have a rough idea what the other threads
are doing at that time. What i would like to look up is how the response
time is changing when everybody else is requesting the same given page.

Is it possible to do something like
- put the whole test in a loop
- add a counter
- replace the numbe of users with the counter value
- store counter value and response time?

Thanks.
SWK


sebb-2 wrote:
>
> You could use the ramp-up time feature to gradually increase the load.
>
> S
> On 11/04/07, SunWuKung <[EMAIL PROTECTED]> wrote:
>>
>> I would like to create a chart that shows the response time of a page
>> depending on the number of concurrent users.
>> I guess I need to first create a thread group with 1 user and run the
>> test,
>> than with 2 users ... and so on.
>> Is there a way to automate this and collect the data?
>>
>> Thanks for the answer.
>> SWK
>> --
>> View this message in context:
>>
http://www.nabble.com/cuncurrency---responsetime-chart-tf3557633.html#a99341
15
>> Sent from the JMeter - User mailing list archive at Nabble.com.
>>
>>
>> ---------------------------------------------------------------------
>> To unsubscribe, e-mail: [EMAIL PROTECTED]
>> For additional commands, e-mail: [EMAIL PROTECTED]
>>
>>
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: [EMAIL PROTECTED]
> For additional commands, e-mail: [EMAIL PROTECTED]
>
>
>

--
View this message in context:
http://www.nabble.com/cuncurrency---responsetime-chart-tf3557633.html#a99397
69
Sent from the JMeter - User mailing list archive at Nabble.com.


---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]




---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]



---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]

Reply via email to