Now first I'd like to clear out any confusion I might have started with the not so clear description, the intention was/is:
first of all to use the MD either as a MCE on a TV or as a desktop environment on the TFT, but have both connected continuously to the MD (although using both at same time seems like a pretty cool idea, I think the setting up wouldn't be worth it and I guess it could get nasty with simultaneous multiple inputs such as keybord/remote) ,
the question because: since I don't have much experience with LINUX I don't really know if (or how if yes) something like this works out of the box, but since I was thinking on UNIX based systems (as for instance in universities) where all clients share the services on one server (here OpenOffice as an example), I thought that it should work without bigger difficulties on LINUX too since its a X system.
The reason for doing this was/is to reduce the cost of the installation having diskless less powerful MDs with the possibility of use as thin-clients (thinking of UNIX systems, I have no idea how it works on Microsofts Media Center Extenders and don't really want to either since I want to get away from MS) and so skip the need of having other PC's to do regular work, and reduce headaches with keeping all systems upgraded and configured and so on. That way I would only need to install the software once on the core and have it available for the whole family (and guests) on all "Media Directed" TV's or desktops.
Now just to see if I got your answers right (please correct if I'm wrong ):
The MDs are running as fully independent OSs using the cores HDD, but is NOT running as an X-window on the core (being only displayed on the MD).
This being so, in order to use client-server architecture I would need to (somehow) start a new (parallel) connection to the core (using either X or VNC as mentioned by dopey) while in the KDE environment of the MD and from there start the desired applications. And as I see it's all do-able due to the X-background of LINUX without the need of any extra HW on the MD side and only depending on the core (the number of, and requirements of the applications that will run simultaneously will determine the core HW). This does not work out of the box and I would need to configure the connection on every MD every time a new MD is installed on the network (not that critical since I won't be installing a new MD every week , although an automated way would be nicer - maybe with some script that creates the connection during boot that only needs to be copied?!).
The other possibility would be to install the desired apps on the MDs since they are fully independent OSs, but then I would need to adapt the HW on the MD to the SW requirements and would still have multiple installations of the same SW, and the SW would be bound to each individual MD.
If I got everything right, I guess I have to start on the reading on server side computing with X and/or VNC on LINUX and so on.
Well, once again thanks everyone, and LinuxMCE is still my main project for the moment. First I'll do some catching up on the reading (I've noticed that asterisk is not so simple either ) and then some smaller testings before I really start on the whole network. In the meanwhile I can get some more ideas for a good overall solution with a nice complete system