All threads are controlled using a WorkManager.
In Servicemix 3.0-M2, this work manager is configured in the
conf/tx.xml
configuration file.
There are a number of parameters that can be configured, but are not
accessible directly from the container.
For example the SEDA queues capacity could be configured using
<sm:container flowName="seda?capacity=100" />
or
<sm:container>
<sm:flows>
<sm:sedaFlow capacity="100" />
</sm:flows>
</sm:container>
However, the queue size for delivery channels can only be configured using
JMX.
You may raise a JIRA for that.
Regarding the threading model, usually JBI components will start their own
thread pool.
However, servicemix components implements the MessageExchangeListener
interface
and use push delivery (instead of the jbi compliant pull mechanism). This
means that
the thread is owned by the container and not the component.
--
Cheers,
Guillaume Nodet
On 7/29/06, anandhs <[EMAIL PROTECTED]> wrote:
Hi Bruce,
We are in the process of evaluating service-mix. For that we plan to run
some performance tests, so I wanted to know what the tuning parameters are
for service-mix. May be I will ask specific questions
Is there anything like tuning service-mix?
How does it decide how many threads to have for NMR?
If the connection between 2 in-vm services is asynchronous then we should
have some thread pool somewhere?
How do I throttle a service from handling more than a certain number of
requests?
With Seda flows there is buffering between services (as I understand it),
is
there an upper limit that can be set on the number buffered?
We have looked a mule and it has a thread pool for each service
(component), is there anything comparable in service-mix?
Thanks for you hellp
Anand
--
View this message in context:
http://www.nabble.com/Tuning-service-mix-tf2017005.html#a5556275
Sent from the ServiceMix - User forum at Nabble.com.
