RE: [Quickfix-users] Quickfix/j performance

[Steve B. <sb...@sm...>]

Hello Lars,
=20
Can you send me a copy of your test program? I've done performance tests =
on
message marshalling and the protocol engine itself (the Session) and it =
was faster=20
than JNI. I'm guessing the performance difference you're seeing is the =
networking code=20
but I'd like to verify that guess.
=20
Thanks,
=20
Steve Bate
Smart Trade Technologies
Phone: +33 4 42 90 03 97
http://www.smart-trade.net/


________________________________

	From: qui...@li... =
[mailto:qui...@li...] On Behalf Of =
lar...@su...
	Sent: Thursday, January 26, 2006 11:27 AM
	To: qui...@li...
	Subject: [Quickfix-users] Quickfix/j performance
=09
=09

	Since I received the error from running with the JNI version of =
quickfix and java I thought it might be nice to test my program with =
quickfix/J. It was almost only to change the libraries. As someone =
mention the quikcfix/j from the initator.getSessions() returns an array =
with the Session(s) objects instead of the SessionID(s). That required a =
small change on just .getSessionID(). Then it worked almost as intended. =
When running the test over one session I get like 1000messages/second =
and with the JNI version I got like 3000messages/second.=20

	=20

	Anyone got any id=E9e why I get lower messages per second using the =
quickfix/J instead of quickfix with JNI?=20

	=20

	The test is very simple it just tries to send x new singe order =
messages as fast as possible over a single session.=20

	I'm using the Session.sendToTarget(message, SessionID); to send =
messages.=20

	=20

	Lars