OSI networking for Windows NT

My interest is in open technologies and networking. Obviously, NT and networking come hand-in-hand, but open technology? How does Windows NT use open technology? The internals of NT are a carefully gaurded secret, but it was written in such a way that various modules can be replaced with third-party components. One important example is NT's networking, which is not dependent on any particular protocol. A user can choose any protocol stack, once it has been written for the NT system. For years, the most-used stack has been TCP/IP, an open techonology. To take this one step further, the OSI protocols could be implemented for Windows NT. My plan is to research what the role of OSI could be in NT's architecture. There are different choices for each of the OSI layers, and my plan is to find which choices are best for NT. This would not be a simple porting issue. The implementation of each layer can be fine-tuned for NT's architecture, to achieve performance, while offering the epitome of platform-independent connectivity, OSI. Windows NT already has a layered architecture, so using the OSI model of networking would fit nicely.

The actual goal of having OSI running on Windows NT is related to NT's new possibility in the role of a router. NT is being used for network routing and bridging already, and companies are investigating using modified versions of NT for high-performance routing, which is usually done by specialized hardware. If the OSI protocols were ported to NT, it could function as a bridge between OSI networks and the more common IP networks. Having this capability without needing specialized bridging hardware would assist OSI's acceptance as a networking protocol in practice. Not to mention the ability to install OSI on one of the most popular PC operating systems.

My interest in the OSI networking model, coupled with my interest in NT as an industry-strength operating system, makes me feel that this investigation will be very interesting and educational. Making OSI available on more platforms will strengthen its reputation as a practical networking solution. After all, who wants a protocol, with no one to talk to?

Tom Larson, larsot@rpi.edu
Rensselaer Polytechnic Institute Computer Science Department