Programming

axes, when you post any code, please add code tags. it makes things much more readable for the people trying to help.

Thanks all, It works for me , I used the following algorithm
If any enhancements, pls correct me...

daemon main function
---------------------
create a socket for listening clients, say socket1
bind the socket
listen to the socket
for ever {
accept connections from clients
create a detachable thread to process the request,
given the socket i.e. socket1
}

thread start_function
--------------------
read the request from given socket i.e. socket1
DO SOME PROCESSING
-----
create another socket for vendor server, say socket2
connect to the socket, socket2
Write received request to socket2
read results from socket2 (vendor server)
close socket2
-----
write final results to client i.e. socket1
close socket1
Exit from the thread

With tags...
daemon main function
---------------------
create a socket for listening clients, say socket1
bind the socket
listen to the socket
for ever {
accept connections from clients
create a detachable thread to process the request,
given the socket i.e. socket1
}

thread start_function
--------------------
read the request from given socket i.e. socket1
DO SOME PROCESSING
-----
create another socket for vendor server, say socket2
connect to the socket, socket2
Write received request to socket2
read results from socket2 (vendor server)
close socket2
-----
write final results to client i.e. socket1
close socket1
Exit from the thread

something wrong with my system.
Even though I insert code tags BUT After posting it is getting distrubed.
Anyway excuse me

I don't know if I'm understanding you correctly, but it seems that the method you are using will make one connection (albiet via a separate thread) to your third-party server per client request? Depending on the volume of requests you incur, they may not like this.

I'd suggest something along the lines of a worker thread model where you create a distinct and known number of worker threads that hang off the third party server. You feed them requests via the worker/consumer queue (see any thread book for guidance on this model) and they then answer the request either via a response queue, or can directly write the results back to the client.

Regardless, this is not the easiest way to implement things. However, constantly spawning threads on a per/request basis may not be the best way to handle things, nor is consuming mass amounts of connections to your third party server.

I suggest picking up a book on thread techniques that will go over various patterns with you and explain how they work. I think the one I have is put out by O'Reilly, but I don't have it with me and I don't remember off-hand.

Also, your main thread could easily handle a resonable amount of client activity by itself without spawning separate threads for each client. I'm not sure how long your client connections last, but keep in mind that many OSes will limit the number of threads you can create in an app. If you have more than that many concurrent clients then you're in trouble and can't spawn any more threads. Keep this in mind if you decide to go that route.