Hola,
I am playing with SwiftMQ, and I have set up an example that looks
pretty much like this:
Point-to-Point
10-25 QueueSessions created on Server w/ accompanying MessageListeners
(seperate machine)
5 QueueRequestors per client (growable pool)
2 clients on different machines (not including server)
100 Mb Ethernet connections on a switched network
the Message Object is not too large, just an xml document with one Element.
Server: processor p3 550MHz 512 MB ram
Clients processor p3 1GHz 512 MB ram
I am only getting a throughput of around 107 per/second for a single client,
and when I add the second client, the througput just about halves for each
client, so still around 100 total, but around 50 for each client. I am
watching the queue, and there are never a large amount of message in the
queue. I have even cached the document, so it is just returned, and not
built each time. Throughput is the same. I have sampled from 100-10,000
messages per client. I do create a individual session for each
consumer/producer, but I don't create a new connection. I am pooling the
QueueRequestors, so I do not have the overhead of creating new Sessions, or
Temporary Queues. Throughput is about the same. In the Consumer/Producer
threads on the JMS Server, I have allocated 150 threads each. Still the
same throughput. None of the machines are underpowered. So what gives.
how can I get more perforamnce?
Any suggestions are greatly appriciated,
Matthew
------------------------------------------------------
SwiftMQ developers mailing list * http://www.swiftmq.com
To unsubscribe from this list, send an eMail to
[EMAIL PROTECTED] and write in the body of your message:
UNSUBSCRIBE developers <your-email-address>
Archive: http://www.mail-archive.com/[email protected]/
winmail.dat
