Can MS succeed in the high performance computing market?

Not that long ago, if you needed a computer for almost anything business-related, you used a proprietary system, usually Unix-based. Along came the PC and personal systems evolved into the machines we all use. In terms of the OS they run, open source advocates say that Microsoft's Windows variants are proprietary and, in the sense that you can't take a software spanner and fiddle around with the equivalent of the pistons and cylinder heads, they're right. But for most practical purposes, Windows defines a standard. And standards are what most people - especially users - are dead keen on.

Today's question is whether Microsoft can push its standard into one of the last redoubts of the proprietary market, high-performance computing. The question is sparked by Microsoft's announcement of Windows Server 2003 High Performance Computing (HPC) Edition, which continues the company's segmentation of the server OS market. Included in this strategy are the Small Business Server, BizTalk Server and ISA Server, to name but a few, all variants on the same theme.

The aim of the HPC Edition according to Microsoft is to make it easier to create Windows clusters, allowing the building of large, scalable systems using multiple low-cost computers. Such systems include large numbers of servers running scientific and technical tasks, such as computational fluid dynamics, weather forecasting, and geology for the oil industry.

What's changed in recent years is the falling cost of hardware, largely as a result of commoditisation and standardisation. While the result is that vendors such as IBM and HP now find it easier to cross-subsidise the cost of hardware out of software revenues, it also lowers the barriers to entry. Effectively, it allows volume players to move into the territory once reserved for the big box, big margin boys.

And as IDC reports, this $2.4 billion market is due to more than double to $5.1 billion by 2008. That makes it very attractive to the likes of Dell, the arch-exponent of commoditised, standardised hardware (except in printer cartridges, strangely enough).

Now Dell already sells workstations and rack servers and partners with EMC to sell storage area networks. It wouldn't take too much to beef up this offering in order to market it as an HPC system. But Dell isn't going to sell a proprietary Unix. It may sell Linux, which we can all agree isn't proprietary and, when (and if?) the HPC version of Windows becomes available, it will surely sell a Windows version, its eyes firmly fixed on capturing some of those high revenues.

And with that step of another corner of computing towards standardisation, hardware prices will fall further and Microsoft's vision will be one step closer to completion.

While Unix vendors such as SCO, HP and IBM on the other hand lick their wounds, Linux vendors will mop up services and support revenue for building those high-end boxes. Of course that also includes bet-hedgers such as IBM and HP (though not SCO), who will have to manage the transition, swapping high revenues for low volume, expensive kit plus a chunky service contract for lower revenues from higher volumes of cheaper kit, plus a chunky service contract. Expect to see a number of new players (in this space) such as Novell and Red Hat, all likely to makes moves in this direction too.

In effect, since all this seems likely to come to pass, Microsoft will have, as it has done so often in the past, helped to validate a market. Hardware vendors will smile, so will Microsoft and Linux vendors. The only ones wearing a frown will be SCO - as long as they lose the court cases - those knew Unix intimately and certainly won't feel at home with Windows, and those who believe in diversity for whatever reason.