Steve Mills on Microsoft
Steve Mills is the head of software at IBM. IBM has five major software "brands" -- Lotus, WebSphere, Tivoli, DB2, and Rational. Recently, he and Bill Gates presented their vision of web services in public, and it's worth reading his take on what's in store for Microsoft next year.
In essence, Steve thinks that Microsoft will try to integrate the client (where it is dominant) with the server (where it is not) and will do this through a combination of closed web-services from the technical side, and useful bundled client/server apps on the marketing side. Predictably, Steve thinks this will fail because 1) the tight bundling is not really that valuable and 2) the Internet has made open standards easier to create and be adopted.
IBM's software group recently re-aligned by industry sector, a move I thought was peculiar at the time because if anything is truly standard across industry, it's the sort of middleware database or management system software that IBM produces. It turns out that this restructuring was there to support building industry specific adaptors and XML schemas into true cross-industry platforms like WebSphere, and so incorporate the industry specific insight IBM has through its services division down into the code it wants to sell into those industries.
I'm a big fan of commoditizing low-level parts of the technology stack so the application builders end up closer to industry needs, and we will see if IBM's approach ends up changing the game to make this more common. Right now, home rolled applications tend to be ugly and buggy and maybe or maybe not close to the real requires specs, while vendor apps (at least Microsoft's) tend to be of extremely high quality (compared to home rolled apps) but quite far from what people actually need to do. I hope that, over time, web services will let home rolled apps be closer to Microsoft quality and specific to the task at hand.
In essence, Steve thinks that Microsoft will try to integrate the client (where it is dominant) with the server (where it is not) and will do this through a combination of closed web-services from the technical side, and useful bundled client/server apps on the marketing side. Predictably, Steve thinks this will fail because 1) the tight bundling is not really that valuable and 2) the Internet has made open standards easier to create and be adopted.
IBM's software group recently re-aligned by industry sector, a move I thought was peculiar at the time because if anything is truly standard across industry, it's the sort of middleware database or management system software that IBM produces. It turns out that this restructuring was there to support building industry specific adaptors and XML schemas into true cross-industry platforms like WebSphere, and so incorporate the industry specific insight IBM has through its services division down into the code it wants to sell into those industries.
I'm a big fan of commoditizing low-level parts of the technology stack so the application builders end up closer to industry needs, and we will see if IBM's approach ends up changing the game to make this more common. Right now, home rolled applications tend to be ugly and buggy and maybe or maybe not close to the real requires specs, while vendor apps (at least Microsoft's) tend to be of extremely high quality (compared to home rolled apps) but quite far from what people actually need to do. I hope that, over time, web services will let home rolled apps be closer to Microsoft quality and specific to the task at hand.
0 Comments:
Post a Comment
Subscribe to Post Comments [Atom]
<< Home