Rafael, > Is it possible that the speedup is coincidental with the publisher finishing?
No I don't think so - the sender was still sending when the receiver got faster after about 15-20 messages. I also tested across two separate machines. Out of interest, I did the same test using Rabbit's experimental STOMP adapter. This did not show the same problem! Best Regards, Carl On 30/04/2008 20:17, "Rafael Schloming" <[EMAIL PROTECTED]> wrote: Gordon Sim wrote: > Carl Bourne wrote: >> As you described it is actually only slower for the first few messages >> - then it speeds up dramatically. Very strange! > > Yes... Rafi, any ideas? Is it possible that the speedup is coincidental with the publisher finishing? I would expect the consumer to be slower than the producer, so at some point the producer should finish and free up resources thereby giving the consumer a speed boost. (This is assuming of course that the producer, consumer, and client are all on one machine.) --Rafael
