Tacvek's Journal: For Those Who Care About Consumer Level Thin-clients

On several occasions I've had those who are somewhat computer literate remark that the current method of managaging several computers in unsatsifactory. They suggested what you and I know of as a thin-client architecture. A home would have one primary computer which might be fairly expensive (although even a cheap consumer PC could handle this). Then throught the house there would be inexpensive thin clients. The advantages of such a system should be clear. There is no need to remember what computer a file is stored on. There is only one computer to update every few years. One only needs to install programs on a single computer etc.

This is not to say that they system is without flaws. Some flaws are obvious. If the main system breaks then no 'system' can be used untill it is repaired. Other problems are more complex. For example, Games generally assume that there is a local graphics processor which outputs directly to the display. Remote hardware accelleration of graphics is currently an under-developed feild. The hardware graphics accellerator shpuld be on the server to keep the costs of the thin-clients down. Games also make other assumptions which might not be true in the case of a thin client.

The biggest problem however is architecural. Windows has some support for such a system, however the support is limited, and is not really intended for such a system. However, if we go with a non-Windows solution, then all the usual problems come into play. All too many people have been tried to use Microsoft Office and would have trouble with the transition to anything else, including OOo. Not to mention the clueless end-consumers would have trouble accepting that the software in the store won't magically work on their systems.

Let us assume that this hypothetical end-consumer has no problems with Microsoft compatability, and while not fully computer literate, happens to have one of us arround to maintain the System. The most obvious way of setting this up is the use of a user-freindly GNU/Linux distribution. The thin clients could be an X terminal. Sounds great. But there are problems. Many GNU/linux appliations that use X11 receive minimal testing of the case where the X11 server is not on the local machine. Some of them have problems with this. The X11 protocol lacks support for sound. Some seperate sound daemon would need to run on the thin client. To the best of my knowledge, there very few such sound daemons still being maintained. AFAIK X11 has no special support for allowing remote clients to access local drives.

One project minimizing the problem with the GNU/linux based solutions is the Linux Terminal Server Project. Perhaps if a solution to the Microsoft Problem(TM) can be found, a thin client platform could work well for end-users. Comments?