Thanks very much everyone. Jerome that's a great link to the docs, thanks! I
can see I have some homework to do in terms of evaluating connectors and
different configurations...
Thanks again!
Aaron
"Jerome Louvel" <[EMAIL PROTECTED]> wrote in message
news:[EMAIL PROTECTED]
Hi all,
Rob is right, the threading strategy depends on the actual server used
(Grizzly, Jetty, Simple or the NRE internal one). In general there is a pool
of worker threads. This pool has a configurable size so you can generally
limit the maximum number of concurrent threads. See this page for
configuration details:
http://www.restlet.org/documentation/1.1/connectors#jetty
For Jetty here is the list of parameters available:
http://www.restlet.org/documentation/1.1/ext/com/noelios/restlet/ext/jetty/J
ettyServerHelper
Now, be aware that web browser generally serialize the request to a single
target server/domain at least limit the number of concurrent requests sent.
If you really want to test the concurrent behavior of your Restlet
application, you should instead use a proper load tester. Here is a
benchmark that was done by Thierry on 1.0 connectors:
http://www.restlet.org/documentation/1.0/benchmark
Best regards,
Jerome
-----Message d'origine-----
De : Rob Heittman [mailto:[EMAIL PROTECTED]
Envoyé : lundi 3 mars 2008 22:45
À : discuss@restlet.tigris.org
Objet : Re: Understanding Restlet's Threading Model?
The mapping of incoming network connections to threads is
very dependent on the HTTP/HTTPS connector/server in use.
Restlet, as far as I know, does not do anything to attenuate
the native behavior of the server with regard to creating
threads for incoming network connections. Which server
environment were you looking at when you tested?
On Mon, Mar 3, 2008 at 2:43 PM, Aaron Crow
<[EMAIL PROTECTED]> wrote:
I'd like to understand the threading model used by my
basic Restlet app. I
have a standalone app that uses Application and
Component, and attaches
subclasses of Restlet to the router. I am using the
reference implementation
provided by noelios. (Many, many thanks to Jerome for
all of this!)