On 22.04.2005, at 17:47, Irving Salisbury III wrote:
> I didn't do that measurement, but I did start running multiple vms and
> have written a load balancer specific to my application. So, I'll
> know more in about 4 hours after we run some performance testing on
> the server. We did some preliminary performance testing and running
> multiple vms instead of multiple threads makes XEP scale vs the
> contention we were seeing inside one VM.
>
Irv,
I've found a four-processor machine for testing, but I don't see
contention at all. It scales to four processors almost linearly. Could
you please make sure that it is not your code (customgenerator or
logger) that causes contention.
I have been testing using CLISER and EnMasse/Actinia; with 1,2,4,8,16
and 32 threads in the server (inside a single JVM). With 4 threads it
runs approximately 3 times faster than with one, and it does not change
its speed when more threads are added (and all threads are loaded in
parallel). I used about 300 files of medium sizes to run through the
engine to test performance.
David
-------------------
(*) To unsubscribe, send a message with words 'unsubscribe xep-support'
in the body of the message to majordomo@renderx.com from the address
you are subscribed from.
(*) By using the Service, you expressly agree to these Terms of Service http://www.renderx.com/tos.html
Received on Fri Apr 22 07:31:49 2005
This archive was generated by hypermail 2.1.8 : Fri Apr 22 2005 - 07:31:50 PDT