The need for a 3D thin clie...

When you talked about workstations some years ago, you talked most of the time of a Sun Workstation. The workstations from Silicon Graphics were an strong competitor. Fueled by an incredible surge in processing power and graphics capability more and more real unix workstations were substituted by workstation-alike systems on x86 technology.

The more i talk with customers in the engineering field i think that this class of workstation alike systems will be only an interlude to something even more lightweight. The models in technical computing get more and more complex. The datasets grows in an incredible speed and thus network bandwidth gets an limiting factor. The problem isn´t the computation in the system. You get dual cpu/dual core systems at an reasonable price, which delivers impressive computational power. But how to get the datasets to this systems. The next problem of this tech pc is, that they store the most valuable asset of engineering company. I heard of to many housebreakings in offices to think about the workstation harddisk as an save place to protect company secrets.

So: Is the deployment of immense compute power to the engineers desktop the correct way? In my opinion the technical PC will more and more start to get a glorified TCP/IP-offloading engine for the graphic card. You need it to visualize the data computed elsewhere. Do you really need a 1000W PC with an own operating system, local storage and so on to do this job? Do you need such a beast at every engineers workplace used only 25% of it´s the time?

I think, the future lies in a engineering thin client. Something like a Sun Ray with a ultrabig fat-ass graphics accelerator. Where is the advantage: As you compute on a server, you are able to connect the data resources with the compute resources with technologies like infiniband or 100 GBit/s-Ethernet, You don´t need to have big pipes to every enginnering workstations. You don´t need to administrate thousands of workstations. You can tightly controll where your engineering data is transported. There is no harddisk in unprotected offices to steal containing the construction blue print of your newest development.

I will end this blog entry with a bold statement: In five years the technical PC is a obsolete technical solution ("Tech PC. Oh, thats so 2006"). Obsoleted by the postulated engineering thin client. Obsoleted by something older: A vastly improved X-Terminal.