You are viewing a plain text version of this content. The canonical link for it is here.
Posted to user@jmeter.apache.org by ziohausam <zi...@gmail.com> on 2009/07/14 12:03:51 UTC
How to distinguish performance drawback source
Hello All,
I have a Question.
I use Jmeter in my Performance test execution.
Actually i always face drawback when number of users increases and of course
this is normal.
what is not normal is that while i check the performance live with the test
execution running, i open a browser session for navigating the web
application and i can't find any drawbacks. also i try this from different
PCs, and no noticeable drawbacks.
For example:
in the testing tool, some request takes 2 seconds with 1 user. and 90
seconds with 75 users running with ramp-up period 40 seconds. that's fine
that's a noticeable drawback. but actually when i use the same PC or any
different PC to call this request though the browser while it is processed
by the test tool, i find no noticeable drawbacks.
How can i explain this? how can i understand this, and also how can i
distinguish performance drawback source (server, application, client machine
hardware or bandwidth, etc)
I had sent this post before in QA forums and had really helpful ideas to
explain the reason, but they still do not represent the real cause as i have
checked all suggestions and found every thing is ok. i hope i can find more
ideas and explanations here too
am
Note: as i am only a beginner using JMETER. most possibly is that i am
missing something configuring my tool or test execution process.
Regards,
Hosam
--
View this message in context: http://www.nabble.com/How-to-distinguish-performance-drawback-source-tp24476707p24476707.html
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
Re: How to distinguish performance drawback source
Posted by sebb <se...@gmail.com>.
On 14/07/2009, Deepak Shetty <sh...@gmail.com> wrote:
> hi
> possible problems that i can think of are
> a. You have loaded the Jmeter client with more load than it can handle (as a
> rule of thumb i wouldn't run more than 20-25 threads from a single client
That very much depends on the test plan (and the host, of course). It
should be possible to run many more threads for many test plans, say
200-300.
One way to check if JMeter is the limiting factor is to compare the
result of running a single instance of JMeter against two instances,
each with half the number of threads.
> jmeter). You would need to distribute your test across multiple machines
> (You can check whether your machine is overloaded by using your OS specific
> perf monitoring tool - reliability and perf monitor on windows)
> b. There is something that you need to tweak in your tests(e.g. if your
> application uses caches , and the browser based tests are hitting the cache,
> whereas the Jmeter tests are not)
> c. You are downloading embedded resources (e.g. big flash/movie files or
> something whereas the browser either has it cached or is streaming it).This
> can easily be tested by checking your script and clearing your browser cache
> before accessing the site while your load test is running.
>
> regards
>
> deepak
>
>
> On Tue, Jul 14, 2009 at 3:03 AM, ziohausam <zi...@gmail.com> wrote:
>
> >
> > Hello All,
> >
> > I have a Question.
> >
> > I use Jmeter in my Performance test execution.
> >
> > Actually i always face drawback when number of users increases and of
> > course
> > this is normal.
> >
> > what is not normal is that while i check the performance live with the test
> > execution running, i open a browser session for navigating the web
> > application and i can't find any drawbacks. also i try this from different
> > PCs, and no noticeable drawbacks.
> >
> > For example:
> > in the testing tool, some request takes 2 seconds with 1 user. and 90
> > seconds with 75 users running with ramp-up period 40 seconds. that's fine
> > that's a noticeable drawback. but actually when i use the same PC or any
> > different PC to call this request though the browser while it is processed
> > by the test tool, i find no noticeable drawbacks.
> >
> > How can i explain this? how can i understand this, and also how can i
> > distinguish performance drawback source (server, application, client
> > machine
> > hardware or bandwidth, etc)
> >
> > I had sent this post before in QA forums and had really helpful ideas to
> > explain the reason, but they still do not represent the real cause as i
> > have
> > checked all suggestions and found every thing is ok. i hope i can find more
> > ideas and explanations here too
> > am
> > Note: as i am only a beginner using JMETER. most possibly is that i am
> > missing something configuring my tool or test execution process.
> >
> >
> > Regards,
> > Hosam
> > --
> > View this message in context:
> > http://www.nabble.com/How-to-distinguish-performance-drawback-source-tp24476707p24476707.html
> > 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
Re: How to distinguish performance drawback source
Posted by Deepak Shetty <sh...@gmail.com>.
hi
There is a HTTP Cache Manager in JMeter but Ive never tried it.
the first thing is , is this your problem though ?
Enable View results tree and see the requests and check what is taking time
in jmeter / what response times you are getting v/s what you see on the
browser (say using firefox and Tamper data/yslow etc)
Also monitor your CPU and memory to see whether 25 is still too much load
per client (how much memory is given to Jmeter as Xmx?)
(in my case I get 20 threads for 2.4Ghz 2GB - 512MB Xmx , but then I have
vista)
regards
deepak
On Wed, Jul 15, 2009 at 6:07 AM, ziohausam <zi...@gmail.com> wrote:
>
>
>
> Deepak Shetty wrote:
> >
> > hi
> > possible problems that i can think of are
> > a. You have loaded the Jmeter client with more load than it can handle
> (as
> > a
> > rule of thumb i wouldn't run more than 20-25 threads from a single client
> > jmeter). You would need to distribute your test across multiple machines
> > (You can check whether your machine is overloaded by using your OS
> > specific
> > perf monitoring tool - reliability and perf monitor on windows)
> > b. There is something that you need to tweak in your tests(e.g. if your
> > application uses caches , and the browser based tests are hitting the
> > cache,
> > whereas the Jmeter tests are not)
> > c. You are downloading embedded resources (e.g. big flash/movie files or
> > something whereas the browser either has it cached or is streaming
> > it).This
> > can easily be tested by checking your script and clearing your browser
> > cache
> > before accessing the site while your load test is running.
> >
> > regards
> > deepak
> >
> >
>
> Thanks for your help
>
> Yes i do simulate only 25 users from1 3 GHz 1GB RAM client.
>
> How can i configure JMETER to use the cached data instead of downloading it
> each time ?
>
> Thanks for your help
>
> Regards,
> Hosam
>
> --
> View this message in context:
> http://www.nabble.com/How-to-distinguish-performance-drawback-source-tp24476707p24497557.html
> 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
>
>
Re: How to distinguish performance drawback source
Posted by ziohausam <zi...@gmail.com>.
Deepak Shetty wrote:
>
> hi
> possible problems that i can think of are
> a. You have loaded the Jmeter client with more load than it can handle (as
> a
> rule of thumb i wouldn't run more than 20-25 threads from a single client
> jmeter). You would need to distribute your test across multiple machines
> (You can check whether your machine is overloaded by using your OS
> specific
> perf monitoring tool - reliability and perf monitor on windows)
> b. There is something that you need to tweak in your tests(e.g. if your
> application uses caches , and the browser based tests are hitting the
> cache,
> whereas the Jmeter tests are not)
> c. You are downloading embedded resources (e.g. big flash/movie files or
> something whereas the browser either has it cached or is streaming
> it).This
> can easily be tested by checking your script and clearing your browser
> cache
> before accessing the site while your load test is running.
>
> regards
> deepak
>
>
Thanks for your help
Yes i do simulate only 25 users from1 3 GHz 1GB RAM client.
How can i configure JMETER to use the cached data instead of downloading it
each time ?
Thanks for your help
Regards,
Hosam
--
View this message in context: http://www.nabble.com/How-to-distinguish-performance-drawback-source-tp24476707p24497557.html
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
Re: How to distinguish performance drawback source
Posted by Deepak Shetty <sh...@gmail.com>.
hi
possible problems that i can think of are
a. You have loaded the Jmeter client with more load than it can handle (as a
rule of thumb i wouldn't run more than 20-25 threads from a single client
jmeter). You would need to distribute your test across multiple machines
(You can check whether your machine is overloaded by using your OS specific
perf monitoring tool - reliability and perf monitor on windows)
b. There is something that you need to tweak in your tests(e.g. if your
application uses caches , and the browser based tests are hitting the cache,
whereas the Jmeter tests are not)
c. You are downloading embedded resources (e.g. big flash/movie files or
something whereas the browser either has it cached or is streaming it).This
can easily be tested by checking your script and clearing your browser cache
before accessing the site while your load test is running.
regards
deepak
On Tue, Jul 14, 2009 at 3:03 AM, ziohausam <zi...@gmail.com> wrote:
>
> Hello All,
>
> I have a Question.
>
> I use Jmeter in my Performance test execution.
>
> Actually i always face drawback when number of users increases and of
> course
> this is normal.
>
> what is not normal is that while i check the performance live with the test
> execution running, i open a browser session for navigating the web
> application and i can't find any drawbacks. also i try this from different
> PCs, and no noticeable drawbacks.
>
> For example:
> in the testing tool, some request takes 2 seconds with 1 user. and 90
> seconds with 75 users running with ramp-up period 40 seconds. that's fine
> that's a noticeable drawback. but actually when i use the same PC or any
> different PC to call this request though the browser while it is processed
> by the test tool, i find no noticeable drawbacks.
>
> How can i explain this? how can i understand this, and also how can i
> distinguish performance drawback source (server, application, client
> machine
> hardware or bandwidth, etc)
>
> I had sent this post before in QA forums and had really helpful ideas to
> explain the reason, but they still do not represent the real cause as i
> have
> checked all suggestions and found every thing is ok. i hope i can find more
> ideas and explanations here too
> am
> Note: as i am only a beginner using JMETER. most possibly is that i am
> missing something configuring my tool or test execution process.
>
>
> Regards,
> Hosam
> --
> View this message in context:
> http://www.nabble.com/How-to-distinguish-performance-drawback-source-tp24476707p24476707.html
> 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
>
>