From: "Luis G. Gómez" Date: 2004-12-30T08:21:44+09:00 Subject: Re: Distributed application Ok, forget the diagram. Picture this: Customer buys a system that is composed of the following hardware: - an app-server - terminal. The app-server contains a script that runs on first startup and: - detects conection to the internet. - connects to a master server (on the internet) and downloads and installs the application upon successfull identification. - Registers and IDs app-server as an active application server. This allows the master server to maintain software remotelly and keep a record of where the app-server is to facilitate locating it from terminals. - starts app-server The terminal contains a script that functions as a thin client providing an interface for authentication and: - detects connection to the internet. - registers with master server and upon corrct id downloads its assigned's app-server information (location, etc...) - Drops connection to master server - locates and connects to assigned app-server - downloads a temporary copy of the application from the app-server upon launch and... - after user is done working removes app from terminal and goes back to "login screen" We know which app-server is "assigned" to which terminal beforehand 'cause we sell it to the customer and maintain record os app-server's IDs and terminal's IDs. We may even forget that part and simply control this with user authentication on a central server (the master server). =========== I'm now thinking customers could request a system that is not maintained remotelly. This means the app-server would be delivered with the latest copy of the application and terminals would have to be configured manually (as far as locating it's app server goes) in the absence of a central "master-server". A connection to the internet would not be necessary in this case. =========== Please excuse me if this sounds disorganized but it's not even in development... I'm barelly brainstorming the idea. Thank you LG