Following is the Jconsole output for
60,000 messages publish by 20 threads, received by 20 threads
mode: in-memory, direct
All ran in local machine, Linux
CPU consumption was high, mem was not accumulated.
[image: Inline image 1]
Need to see who use CPU.
Thanks
On Sun, Oct 12, 2014 at 7:47 PM, Hasitha Hiranya hasit...@wso2.com wrote:
Hi MB Team,
Problems we had previously with OnFlightMessageTracker class (class
responsible for tracking all message deliveries)
OnFlightMessageTracker had many complex methods (not grouped well to
methods)
It had AMQP specific stuff
It had time based checks
It had synchronization issues which dropped TPS
if we track messages fromTestAndAdd we had problems
I happened to implement a new one and got rid of all above during
week-end.
But this needs to be tested well. I have tested with in-memory direct mode
and it works well. Got a TPS of 2500 from 20 threads
(For JDBC we need to do order by in JDBC level itself. Collection.sort()
after receiving results did not work.)
I vote replacing this with existing. Will we discuss this implementation
tomorrow morning?
Thanks
--
*Hasitha Abeykoon*
Senior Software Engineer; WSO2, Inc.; http://wso2.com
*cell:* *+94 719363063*
*blog: **abeykoon.blogspot.com* http://abeykoon.blogspot.com
--
*Hasitha Abeykoon*
Senior Software Engineer; WSO2, Inc.; http://wso2.com
*cell:* *+94 719363063*
*blog: **abeykoon.blogspot.com* http://abeykoon.blogspot.com
___
Dev mailing list
Dev@wso2.org
http://wso2.org/cgi-bin/mailman/listinfo/dev
