Re: Performance compared to official performance statement

[maxant <as...@maxant.co.uk>]

Hi all,

I too cannot get near the figures quoted.

I have an Intel i5 2.4 GHz dual core.  I run the ActiveMQ Server and the
unit test which pumps messages into the server, localhost (so no network) -
see the code at
https://github.com/maxant/share/blob/master/DBJMSPerf/src/JMSWriter.java

I only manage about 200 messages a second (< 5 milliseconds per message).
The CPU isn't particularly active but disk activity seems fairly high (based
on the light flashing).

In my case, I handle transactions myself, because I want to simulate
committing after each message and I re-use the connection since I want to
simulate connection pooling.

I use the persistent delivery mode because I am trying to analyse a design
which has high availability - messages may not be lost.

Is the code / configuration used in the cited performance tests available
anywhere?
How could I improve performance in my setup?  

If I make write messages non-transactionally and non-persistently then
performance improves, but I'm still only getting 300 messages a second.  

The messages are about 1K big.

Thanks for any tips / pointers.
Ant




--
View this message in context: http://activemq.2283324.n4.nabble.com/Performance-compared-to-official-performance-statement-tp2354410p4675880.html
Sent from the ActiveMQ - User mailing list archive at Nabble.com.