You have to feel sorry for Sun's chief executive Scott McNealy. In some ways, he is someone who almost got everything right - the central role of Unix, the shift towards networked computing, the increasing importance of platform-independent software - and yet has failed to reap the rewards of his acuity.
This track record means that the major strategic initiatives of his company are always worth looking at, and the latest is no exception. Its announcement formed the heart of the SunNetwork 2002 conference. There is an accompanying press release and company feature on the subject.
The rather uninspired name for Sun's new vision is N1; it has a home page, a short FAQ and a white paper. There are also some essays, though with the honourable exception of Yousef Khalidi's, these are frankly rather thin.
The contrast between the fanfare that accompanied the project's announcement and the paucity of detailed information about it makes N1 a vapourware announcement that even Microsoft would be proud of. Nonetheless, it seems clear that Sun's great N1 breakthrough - what it rather pretentiously describes as "unifying all the resources in a compute fabric" - amounts to little more than a re-invention of the grid services model discussed in this column six months ago.
That is, N1 seems to be about allowing all kinds of computing resources - processing, storage, etc - to be allocated in a dynamic and seamless way to tasks as and when they arise. Done properly, this would allow a company's IT infrastructure to be utilised far more efficiently - one of the driving forces behind the N1 project - as well as providing for easy scalability, which is an important consideration in the context of e-commerce.
But Sun arrives at this particular party rather late. Grids have been around for a number of years, and are already entering the mainstream. Moreover, a big question mark hangs over N1: the extent to which it will embrace rival platforms. Sun's chief technology officer Greg Papadopoulos talks of "open architecture", but this can mean anything, as Microsoft's frequent invocation of the phrase demonstrates.
The openness of the N1 architecture is crucial. If it is not completely platform-independent, as grid services are, then Sun is merely offering a proprietary version, with all the unacceptable lock-in that this implies. However, if it is truly open, Sun will be faced with the fact that users will inevitably opt for the best-value back-end platform on which to run the N1 superstructure. And that means GNU/Linux, as just about every large-scale grid project shows.
If N1 is to stand a chance, Sun must fully confront the GNU/Linux challenge that so far it has only tiptoed around. There are already signs that Sun is seeking some kind of accommodation with this rising star. After acquiring Cobalt in 2000 without even mentioning the dreaded L word, it has recently come out with a new GNU/Linux-based server, LX50, and even - heresy of heresies - something called
But with N1, such half-measures will not be enough. Sun must embrace GNU/ Linux as whole-heartedly as IBM: if it doesn't, N1 will either flop or be run on largely non-Sun hardware and software, with all that this implies for the company. Let's hope McNealy gets it completely right this time.