Re: Maintaining near constant TPS

If you run your tests with less threads, is the TPS more consistent or does
it continue to have the large variation?

Have you tried running multiple worker processes with fewer threads ( 4
workers with 50 threads)?  How about a single agent with 200 threads
(literally, stop the other for agents)?

Is the initialSleepTime property set?  This should be in milliseconds.

Are you doing response processing?  We process a lot of JSON and ran into
major issues with SimpleJSON and org.json.  If you are using either of
these, it would be well worth the effort to move to jackson, gson or some
other high performance json parser.

What OS are your agents?  If you are running linux, you really want to make
sure your average load is <= the number of cores.

There are many things you can try.  Calibration and tuning is an import
part of the script creation...

Hope this helps,

-Marc




On Wed, Feb 5, 2014 at 5:30 AM, Gary Mulder <[email protected]> wrote:

> So you're seeing no Full GC's in the Grinder's GC log?
>
> Gary
>
>
> ------------------------------------------------------------------------------
> Managing the Performance of Cloud-Based Applications
> Take advantage of what the Cloud has to offer - Avoid Common Pitfalls.
> Read the Whitepaper.
>
> http://pubads.g.doubleclick.net/gampad/clk?id=121051231&iu=/4140/ostg.clktrk
> _______________________________________________
> grinder-use mailing list
> [email protected]
> https://lists.sourceforge.net/lists/listinfo/grinder-use
>
>

------------------------------------------------------------------------------
Managing the Performance of Cloud-Based Applications
Take advantage of what the Cloud has to offer - Avoid Common Pitfalls.
Read the Whitepaper.
http://pubads.g.doubleclick.net/gampad/clk?id=121051231&iu=/4140/ostg.clktrk

_______________________________________________
grinder-use mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/grinder-use