> First performance test - I pushed 1 million messages through in 1,108
> seconds. This averages to throughput of just under 1,000 messages per
> second.
>
> This is a little sluggish, but it is adequate performance for most
> production systems.

ActiveMQ that uses embedded Derby on Xeon 3,02 GHz, SATA HDD, Sun JDK 1.4.2,
Gentoo Linux gives me between 8,000 and 15,000 messages per second. That is
almost enough for my needs (though they are quite specific).

For comparison, JGroups give on the same hardware something around 1,000
messages using IP multicast and ~2,000 messages on TCP stack.

> Revamped in C++ and tied directly to the Firebird buffering layer, I
> expect significant improvement because of elimination of duplicated
> work.

Try JayBird with embedded engine - you will know more or less the raw
performance you can get. Usually it is twice as fast as accessing database
over the wire.

> Firebird used steady 60%+/- 2% of CPU throughout running test case. JVM
> that included both the queues and the testcases used about 35% +/- of
> CPU throughout the running of the test case.

There is unneeded overhead of the TCP layer.

Roman