Being Fast, Cheap and Flexible

Ask Motorolas engineers what they did on their summer vacation. The day after Labor Day, the company announced it had succeeded in growing layers of high- performance gallium arsenide on a foundation wafer of less expensive, physically tougher silicon.

More than a cool science project, this development promises lower costs for optical communication hardware, as well as affordable microchips at speeds in the tens of gigahertz.

The problem with growing GaAs on silicon is that the crystal structures are different, but adjacent layers want to get along. In the process of trying to adjust to each other, the layers develop cracks. The ingenious solution, which has spurred Motorolas application for 270 patents, involved the development of a material that could bond to both types of crystal as an intermediate layer.

You know whats coming, dont you. I can never resist a good metaphor. When I read about Motorolas development, it reminded me of what Microsoft is trying to do with .Net; what Microsofts own representatives agree with me has not been well communicated. Let me try.

Within an all-Windows environment, Microsofts COM has worked pretty well, but it hasnt done a good job of crossing boundaries into the mixed architectures of the Web. The company wants to preserve the rich interaction of the Windows desktop, while still being sufficiently platform-neutral to compete with technologies such as Java. In short (ahem), the company needs a way to grow a high-performance layer on a flexible foundation.

.Net lets an application developer use the facilities of Windows, including newly unified programming models for teams using more than one programming language, while interacting with non-Windows systems via open protocols.

Both Motorolas chips and Microsofts Web services APIs demonstrate the need to focus on finding ways to make things work together. Things today, in isolation, work pretty well; success tomorrow lies in superior integration.