MS planning shedloads of Windows NT variants

Weird noises from this week's Gartner Symposium in San Diego say Microsoft, despite apparently postponing development of one version of NT, is planning something in excess of ten different new versions. But from the reports it seems clear that the scheme is going to fall awkwardly between the techie plan, which is converging on one code base and then tailoring in different directions, and the marketing one, which is give us all your money. Marketing will as always triumph in the definition of MS strategic direction. Hence the planning as reported by Gartner presumes the leap of faith that says Microsoft is going to get a version of Windows 2000 that it can build on out of the door by 2000. The tally of NT (or Win2k if you like) versions that will then start to appear runs approximately as follows: We already have three, in the shape of Workstation, Server and Enterprise Edition. You might say that calling these three NT-based operating systems rather than just the one operating system is stretching packaging a little now, but there's plenty more where that came from. In addition to these there will be new embedded and data centre versions, and then smart card and CE versions (look, read the rest of the article before you flame us -- we're just reporting, OK?). These different versions will be multiplied further by the move to 64-bit. This will happen at the top end first, data centre and enterprise by 2001 (sure...) but everything else will follow. More versionitis comes in the shape of projected splits in the number of processors that different versions will support, so four way, eight way and beyond implementations of Win2k will be presented as being more powerful, and therefore will cost lots more dollars. The consumer release of NT, incidentally, is now targeted for 2002. This explains the recent retrieval of Windows 98 from the corporate morgue -- jump leads are being applied to the cold corpse of Good Old 9x even as you read. We can presume that Gartner's roadmap come from the horse's ass, er, mouth -- Redmond itself, and that they were presented as a series of technological innovations that marked the move away from a 'one size fits all' strategy. But the truth is clearly the reverse. If the market needs lots of different operating systems for lots of different platforms, then obviously there's no point in them all being NT, is there? And if you look at it properly, they inevitably won't all be NT. The core workstation and server versions will be, certainly, and the various 32/64-bit and scalability permutations will be NT as well. But clear away all the smoke from that bit of the roadmap and what you get is Microsoft with two variants of the OS which move first to Win2k and then to -bit (not necessarily as smoothly as the marketing folk would like, of course). You still get two variants, and the enterprise and data centre stuff just means pay Microsoft more money to get access to the functionality that's in there in the first place. And before you write, yes we do know we're probably being overly generous in counting the current product line as two, given that the only thing stopping you running Web servers on Workstation is the licence. The "embedded" version is currently another one that owes more to packaging than technology. Right now Microsoft has a few customers who're operating point of sale equipment running NT, but this is not embedded software as others would understand it -- what you get is an NT PC, some of whose bits you can't get at. The smart card and CE versions are weirder, and therefore further down the line. Win2k obviously doesn't fit on a smart card, and presumably anything MS builds that fits on a smart card and is called NT is going to be a different operating system, ie. another branding exercise. And NT for CE is even more bizarre -- an operating system for an operating system? No, this is a muddled indication that the convergence plan is still twitching -- a "version" of NT will be built to take over from the current line of CE operating systems, and Microsoft will no doubt confuse itself further by remembering projected games/consumer/set-top box variants of CE and then bifurcating NT some more. Basically, friends, it's all a complete crock. Even if Microsoft does manage to get a stable Win2k it can build on out of the door, the 'building' process will largely degenerate into a combination of packaging-differentiated products and entirely different operating systems held together, sort of, by being called NT (or whatever it's called by then). But then, it's a bit like that already, isn't it? ®