Hi Tharindu,
Tharindu Mathew wrote:
Hi,
I subscribed to the labs mailing list and forwarded our conversation there.
Shall we talk there on the list?
Yes, perfect, thank you.
answered the questions, inline... Please correct me if I'm wrong.
Regards,
Tharindu
On Tue, Mar 10, 2009 at 11:23 PM, Bernd Fondermann <[email protected]>wrote:
Hi,
I tried to clarify the proposal text.
Vysper documentation is here:
http://cwiki.apache.org/labs/vysper.html
and the source code and all the rest is here...
http://svn.apache.org/repos/asf/labs/vysper/
If you have problems getting it running, please ask.
I will get it running and get back to you asap.
more comments below, inline...
Tharindu Mathew wrote:
Regards,
Tharindu
Tharindu Mathew wrote:
Hi,
I'm a student interested in the Vysper project. I have implemented a
server
using asynchronous I/O (Using Java NIO) but the implementation wasn't
very
stable.
Cool. Vysper uses MINA, so it also benefits from NIO features.
Yes, I heard of the MINA framework, didn't use it though. As I said it was a
basic implementation, nothing heavy duty :)
I was wonderin what would be the scope of this server.
In the end, you decide :-) - I'd suggest to concentrate on implementing
pubsub.
I guess that would be following the spec for pubsub. So handling issues of
the server is not within the scope?
Well, if you find issues with the server or have to extend/change it at
places to complete your task, this is well within the scope.
Would it be
implemented to the extent of handling load balancing, clustering, etc. Or
would a simple implementation do? (which would probably need a lot of
work
to go into an actual production use)
The Vysper server is not ready for real heavy duty - that would be a
different project altogether.
What would be your production usage scenario?
I was thinking this was an enterprise ready solution and making it
enterprise ready was part of the scope. I thought of a production scenario
like a scaled down version of Google Talk (around 1000 users).
I think 1000 users are a realistic goal. Please be aware that some XMPP
server installation scale into the millions-of-users scenarios. So what
"enterprise ready" really means might vary heavily betweenn enterprises ;-)
Much has been said about scaling problems of this kind of services,
especially regarding to twitter's downtimes. The problem is
pubsub-immanent and I think the spec has broad enough that scaling
should really be a secondary goal.
Bernd
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
