jmeter-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Oliver Lloyd <oliver_ll...@hotmail.com>
Subject Re: Jmeter Performance using, jmeter-server VS running in the local instance
Date Wed, 21 Sep 2011 18:14:28 GMT
Ah, that's different. It's true (or at least I have experienced the same
thing) that if you have JMeter running in Distributed (master_slave) mode
then you can potentially hit an IO bottleneck where JMeter cannot write
multiple results streams to one file quick enough. The workaround - as you
have already seen - is to run multiple instances in isolation and then
append all the results together (and then sort them). Personally, I do this
all the time - I have it as a habit.

I assume this is an IO issue, might not be, but it is a problem where
multiple processes are trying to write to one file. It's not slow by any
means, but at some point there is a limit.

You can play around with the config options to get around this
(HOLD||BATCH||STATISTICAL) should all help but each has it's price.

As an aside, all load testing tools have the same problem. LoadRunner solves
this by only sending back aggregated summary data as the test is running and
then at the end you have to wait while it compiles the raw data - for large
tests this can take an age - This is effectively the same as selecting a
hybrid of BATCH & HOLD mode in JMeter (kind of).

PS. That is all independent of using the CTT. That's just useful for
creating repeatability / reaching a defined load. etc.

-----
http://www.http503.com/
--
View this message in context: http://jmeter.512774.n5.nabble.com/Jmeter-Performance-using-jmeter-server-VS-running-in-the-local-instance-tp4631144p4827319.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


Mime
View raw message