jmeter-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From moon <moonmun...@yahoo.co.in>
Subject Re: How throughput is calculated in jmeter?
Date Thu, 15 Nov 2007 11:56:14 GMT

OK I go with yo....Plz explain in detail about the calculations here.....

As yo said Login requests(with 1 sample) seems to be fine.....but see the
send and receive
and further requests(which has 5 samples) and how is the throughput
calculated there...

for login requests, approximately it takes 27 sec(from average) for 1
 requests, hence the throughput 2.2/min....it seems to be fine....
But for other requests,
ex..sendreceive requests, approximately it takes 49 sec(from average) for 1
requests, here how we get throughput value to be 35.8/hr...



As I wrote previously, the Aggregate Report is wrong on two counts:

* KB/sec is actually bytes/sec
* Throughput calculation is incorrect (see previous post for full details).

Use the Summary Report instead.

On 15/11/2007, moon <moonmunees@yahoo.co.in> wrote:
>
> As yo said Login requests seems to be fine.....but see the send and
> receive
> and further requests and how is the throughput calculated there...
>
> for login requests, approximately it takes 27 sec(from average) for 1
> requests, hence the throughput 2.2/min....it seems to be fine....
> But for the next requests,
> ex..sendreceive requests, approximately it takes 49 sec(from average) for
> 1
> requests, here how we get throughput value to be 35.8/hr...
>
> Whether KB/sec column in aggregate graph is wrong? In summary report
> whether
> KB/sec values are calculated correctly?
>
> Plz explain me in detail............................(version of the Jmeter
> used by me is 2.3)
>
>
>
>
>
>
> sebb-2-2 wrote:
> >
> > On 14/11/2007, sebb <sebbaz@gmail.com> wrote:
> >> On 14/11/2007, sebb <sebbaz@gmail.com> wrote:
> >> > On 14/11/2007, moon <moonmunees@yahoo.co.in> wrote:
> >> > >
> >> > > Hi,
> >> > >
> >> > > I ran a test plan with the following samplers(HTTP requests) under
> a
> >> single
> >> > > thread.
> >> > > 1.Login requests(under once only controller)
> >> > > 2.Sendreceive requests
> >> > > 3.More requests
> >> > > 4.Image requests
> >> > > 5.Next requests
> >> > > 6.Previous requests
> >> > >
> >> > > the above test plan is executed with,
> >> > >
> >> > > Number of threads-1
> >> > > Ramp-up-0 sec
> >> > > loops-5
> >> > >
> >> > > and test ran time is 589000 milliseconds...
> >> >
> >> > Which version of JMeter are you using?
> >> >
> >> > > I got the following aggregate report,
> >> > >
> >> > > http://www.nabble.com/file/p13746933/ScreenHunter_124.gif
> >> > >
> >> > > Summary report
> >> > >
> >> > > http://www.nabble.com/file/p13746933/ScreenHunter_125.gif
> >> > >
> >> > > why the throughput in aggregate report and summary report differs?
> >> and how
> >> > > is the throughput calculated.
> >> >
> >> > Throughput = number of requests/elapsed time.
> >> >
> >> > The numbers of requests are identical (as are the other columns apart
> >> > from KB/sec), so it looks like the two Listeners are using slightly
> >> > different ways of determining the elapsed time.
> >> >
> >> > It's possible that one of the Listeners is assuming that the sample
> >> > timeStamps are end timeStamps and the other start stamps.
> >
> > I've found the problem - the statistics calculator used by the
> > Aggregate Report was mistakenly adding the elapsed time to the end
> > time; this had the effect of increasing the calculated elapsed time,
> > and therefore reducing the throughput.
> >
> > The fix will be in the next release of JMeter.
> >
> > The Summary Report has the correct calculation; it also uses fewer
> > resources as it does not need to save all the samples.
> >
> > [It's best not to use the Aggregate Report unless you really need the
> > Median and 90% values.]
> >
> > Although the Aggregate Report shows a lower throughput, the
> > discrepancy will be small, especially for longer test runs, where
> > individual elapsed times are a small percentage of the overall run
> > time.
> >
> > It was more obvious in this case because there were few samples - e.g.
> > only 1 login request; the Aggregate throughput was half the Summary
> > throughput because the elapsed time was used twice by the Aggregate
> > calculation. The TOTAL lines are closer - 2.6 against 2.7 which is
> > about 4% difference.
> >
> > Thanks for reporting the problem.
> >
> >> > [The Aggregate report looks to have the wrong heading; should
> probably
> >> > be bytes/sec. Or the code failed to divide by the appropriate
> factor.]
> >>
> >> The Aggregate Graph and Aggregate Report code failed to divide by
> >> 1024; this has been fixed for the next release.
> >>
> >> > If you have a copy of the sampler data (JTL file) could you send it
> to
> >> > me privately please? Either CSV or XML format will do. [Please don't
> >> > post it to the list]
> >> >
> >> > > Would you explain in detail the throughput calculation for each
> >> requests
> >> > > with formula .....plz...
> >> >
> >> > See above.
> >> >
> >> > > --
> >> > > View this message in context:
> >>
> http://www.nabble.com/How-throughput-is-calculated-in-jmeter--tf4805121.html#a13746933
> >> > > Sent from the JMeter - User mailing list archive at Nabble.com.
> >> > >
> >> > >
> >> > >
> ---------------------------------------------------------------------
> >> > > To unsubscribe, e-mail: jmeter-user-unsubscribe@jakarta.apache.org
> >> > > For additional commands, e-mail:
> jmeter-user-help@jakarta.apache.org
> >> > >
> >> > >
> >> >
> >>
> >
> > ---------------------------------------------------------------------
> > To unsubscribe, e-mail: jmeter-user-unsubscribe@jakarta.apache.org
> > For additional commands, e-mail: jmeter-user-help@jakarta.apache.org
> >
> >
> >
>
> --
> View this message in context:
> http://www.nabble.com/How-throughput-is-calculated-in-jmeter--tf4805121.html#a13761123
> Sent from the JMeter - User mailing list archive at Nabble.com.
>
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: jmeter-user-unsubscribe@jakarta.apache.org
> For additional commands, e-mail: jmeter-user-help@jakarta.apache.org
>
>

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




-- 
View this message in context: http://www.nabble.com/How-throughput-is-calculated-in-jmeter--tf4805121.html#a13765803
Sent from the JMeter - User mailing list archive at Nabble.com.


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


Mime
View raw message