jmeter-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From sebb <seb...@gmail.com>
Subject Re: Performance of if controller
Date Sun, 07 Sep 2008 11:41:19 GMT
On 07/09/2008, Ian Blavins <iblavins@temenos.com> wrote:
> G'day
>
>  The If only always evaluated to true while I was testing its performance.

So to establish whether or not the If Controller is more inefficient
than other controllers, one needs to compare if with another
controller that behaves the same - which you have now done.

> In real usage it would evaluate to true or false dynamically depending on other conditions
in the iteration.
>
>  No words missing. I was saying that I reimplemeted the If controller with a switch controller
but as I was at 2.3.1 that was doomed to failure. After I upgraded to 2.3.2 the switch controller
worked fine. And its performance is way better than the If so that is the structure I will
use going forward.
>

OK.

>
>
>  Ian Blavins
>  Software performance specialist
>  .
>  TEMENOS
>  The Banking Software Company
>  .
>  PeopleBuilding 2, Maylands Av
>  Hemel Hempstead   UK   HP2 4NW
>  .
>  T:  +44 (0) 1442 431 106
>  E:  iblavins@temenos.com
>  .
>  www.temenos.com
>  .
>   Consider the environment. Please don't print this e-mail unless you really need
to.
>
>  -----Original Message-----
>  From: sebb [mailto:sebbaz@gmail.com]
>
> Sent: Saturday, 6 September 2008 10:03 PM
>  To: JMeter Users List
>  Subject: Re: Performance of if controller
>
>  On 06/09/2008, Ian Blavins <iblavins@temenos.com> wrote:
>  > G'day
>  >
>  >  600 threads
>  >
>  >  Tried a loop controller using the output of the jmeter sampler that determined
whether to execute the script or not. The output was 0 or 1 which went into the loop count
field of the loop controller. Didn't get stable behavior.
>
>  Actually, I meant that a loop controller with a fixed count of one
>  should be equivalent to a Simple Controller and an If Controller where
>  the condition is always true.
>
>  In all cases, the samples are executed once.
>
>  >  Replaced the loop controller with a switch controller but that didn't work because
it doesn't at 2.3.1.
>
>  Some words seem to be missing here?
>
>  >  Upgraded to 2.3.2 and at least I now have a working structure. Next is to see
if the performance is any better.
>  >
>  >
>  >
>  >
>  >
>  >  Ian Blavins
>  >  Software performance specialist
>  >  .
>  >  TEMENOS
>  >  The Banking Software Company
>  >  .
>  >  PeopleBuilding 2, Maylands Av
>  >  Hemel Hempstead   UK   HP2 4NW
>  >  .
>  >  T:  +44 (0) 1442 431 106
>  >  E:  iblavins@temenos.com
>  >  .
>  >  www.temenos.com
>  >
>  > .
>  >   Consider the environment. Please don't print this e-mail unless you really
need to.
>  >
>  >
>  >  -----Original Message-----
>  >  From: sebb [mailto:sebbaz@gmail.com]
>  >  Sent: Saturday, 6 September 2008 4:36 PM
>  >  To: JMeter Users List
>  >  Subject: Re: Performance of if controller
>  >
>  >  On 06/09/2008, Ian Blavins <iblavins@temenos.com> wrote:
>  >  > G'day
>  >  >
>  >  >
>  >  >
>  >  >  I've been trying to identify the cause of a major loss of throughput in
>  >  >  my JMeter scripts (note, in the throughput of the scripts through the
>  >  >  JMeter engine, not loss of throughput on the target server).
>  >  >
>  >  >
>  >  >
>  >  >  It appears the If controller is very expensive in JMeter throughput
>  >  >  terms.
>  >  >
>  >
>  >  Could well be so - it has to create a Javascript interpreter.
>  >
>  >  >
>  >  >  If I wrap the sampler call of an existing test plan with an If
>  >  >  controller that has an expression of "true" == "true" then the
>  >  >  throughput of my scripts drops from around 36,000 scripts per second to
>  >  >  around 2,200 scripts per second.
>  >  >
>  >
>  >  How many threads?
>  >
>  >  >
>  >  >  This is jmeter 2.3.1 with the client on Windows XP and remote jmeter
>  >  >  hosts on an IBM p570 running AIX.
>  >  >
>  >
>  >  Current version is 2.3.2
>  >
>  >  >
>  >  >
>  >  >  Is there a known slowness in the If controller ?
>  >  >
>  >
>  >  What happens if you replace the If Controller with a simple controller
>  >  or a loop controller with count=1?
>  >
>  >  That would be a fairer comparison.
>  >
>  >  >
>  >  >
>  >  >
>  >  >  (These are test scripts designed to determine how to get high JMeter
>  >  >  throughput so they don't do very much. In my real scripts half the
>  >  >  expression is of course a variable computed in earlier test plan steps.)
>  >  >
>  >  >
>  >  >
>  >  >  I can easily get around this since my sampler is a custom written Java
>  >  >  sampler. I just have to pass the test variable into the sampler and let
>  >  >  it decide whether to execute.
>  >  >
>  >  >
>  >  >
>  >  >
>  >  >
>  >  >
>  >  >
>  >  >  Ian Blavins
>  >  >  Software performance specialist
>  >  >
>  >  >  .
>  >  >
>  >  >  TEMENOS
>  >  >  The Banking Software Company
>  >  >
>  >  >  .
>  >  >
>  >  >  PeopleBuilding 2, Maylands Av
>  >  >
>  >  >  Hemel Hempstead   UK   HP2 4NW
>  >  >
>  >  >  .
>  >  >
>  >  >  T:  +44 (0) 1442 431 106
>  >  >  E:  iblavins@temenos.com
>  >  >
>  >  >  .
>  >  >
>  >  >  www.temenos.com <http://www.temenos.com>
>  >  >
>  >  >  .
>  >  >
>  >  >  P Consider the environment. Please don't print this e-mail unless you
>  >  >  really need to.
>  >  >
>  >  >
>  >  >
>  >  >
>  >  >  Disclaimer:
>  >  >  If you have received this e-mail in error please notify the sender.
>  >  >  Please note that any views or opinions presented in this e-mail are solely
>  >  >  those of the author and do not necessarily represent those of TEMENOS.
>  >  >  We recommend that you check this e-mail and any attachments against viruses.
>  >  >  TEMENOS accepts no liability for any damage caused by any malicious code
>  >  >  or virus transmitted by this e-mail.
>  >  >
>  >  >
>  >
>  >
>  > ---------------------------------------------------------------------
>  >  To unsubscribe, e-mail: jmeter-user-unsubscribe@jakarta.apache.org
>  >  For additional commands, e-mail: jmeter-user-help@jakarta.apache.org
>  >
>  >
>  >
>  >  Disclaimer:
>  >  If you have received this e-mail in error please notify the sender.
>  >  Please note that any views or opinions presented in this e-mail are solely
>  >  those of the author and do not necessarily represent those of TEMENOS.
>  >  We recommend that you check this e-mail and any attachments against viruses.
>  >  TEMENOS accepts no liability for any damage caused by any malicious code
>  >  or virus transmitted by this e-mail.
>  >
>  >
>
>  Disclaimer:
>  If you have received this e-mail in error please notify the sender.
>  Please note that any views or opinions presented in this e-mail are solely
>  those of the author and do not necessarily represent those of TEMENOS.
>  We recommend that you check this e-mail and any attachments against viruses.
>  TEMENOS accepts no liability for any damage caused by any malicious code
>  or virus transmitted by this e-mail.
>
>
Mime
View raw message