Tracy, I think you're going to find the latter of your two approaches to be more efficient and quite easy to accomplish, especially the server will be authored in .NET. If you're using the term port to refer to the actual socket, then you'll find that handling multiple clients concurrently is fairly easy to achieve without needing to filter results on the client side.
The main issues you're going to find with this sort of thing (network / tunneling issues, thread management) will probably be non-existent in a controlled environment with 100 clients. I see the very basic architecture as something like this: Flash client <-> xml/ascii serialization/deserialization layer <-> .net socket server <-> xml/ascii serialization/deserialization layer <-> system internals or .net remoting to the service aggregator / whatever The other option in which you could avoid writing a server would be to poll (even though it seems yucky) very quickly. Probably not such a huge burden if you're on a lan and use flash remoting. Jim -----Original Message----- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of Tracy Spratt Sent: Friday, November 04, 2005 2:44 PM To: [email protected] Subject: [Flashcoders] Need advice on XMLSocket architecture Hello, I'm a regular on the flexcoders list, where it was suggested that I ask my question over here. Any advice will be welcome! I have 100 fixed, identifiable clients running a Flex 1.5 app on an intranet. A Dot.Net integration tier maintains individual states for each client. The integration tier talks asynchronously to third party apps over async tcp sockets and even mail transport. I had planned on a leisurely polling setup to get the clients updated with changes in state, but I now have a requirement for an update with minimum latency. The messaging between clients and server is very low frequency (a few clients updated per minute) and very low message size (a few k per update). I am now looking at a true "push" connection using XMLSockets. I see two approaches: On would be to have a separate port connection for each client with each client only getting their messages, the other would be a single port, where all clients would get all messages, and ignore those that were not theirs. In my case performance, reliability and simplicity are most important, and scalability not so much an issue. Any suggestions, experiences, resources would be appreciated!! Tracy _______________________________________________ Flashcoders mailing list [email protected] http://chattyfig.figleaf.com/mailman/listinfo/flashcoders _______________________________________________ Flashcoders mailing list [email protected] http://chattyfig.figleaf.com/mailman/listinfo/flashcoders
