Hello,
Trying to improve the performance on our application, i tried to replace the plumbering -
as i saw the performance should be signifantly better with JBossMessaging.
We use the JBoss4.0.4GA .
I followed the How-to wiki and had to change some lines of code (I should open a minor bug
on the queue.send function ) and achieved functional tests.
But when i tried to stress my application with our homemade stress tool l was very
deceived - all was very slow.
I would like to know if our architecture is not responsible of this poor performance :
We use 1 queue with many selector (20 in my test) based on a hash on the selector
identifier. We have so 20 sequencer scaning for message in one queue.
I read this is a not-performant way of processing message (JBMessaging-275) - but is it
normal that we have a so obvious difference with JBossMQ ?
Olivier
View the original post :
http://www.jboss.com/index.html?module=bb&op=viewtopic&p=3973811#...
Reply to the post :
http://www.jboss.com/index.html?module=bb&op=posting&mode=reply&a...