Platform games are serious business

For the first time since glam rock, platforms are in. If they're not a good fit, expect to stumble
Written by Leader , Contributor

PCs have never been the perfect business platform. Unlike their mainframe and minicomputer forebears, they were never designed for central control. That's cost the sanity of many a brave IT professional.

Intel wants to put that right with its forthcoming professional business platform. Although the company hasn't said exactly what will be included in next month's launch, we'll bet on nearly all of the Intel Ts: AMT for management, VT for virtualisation, LT for security, EM64T for 64-bittedness (but not HT, which is going down with the Netburst ship).

It's about time. Although the modern PC client looks very different to its 1981 IBM PC ancestor, there have been remarkably few pure enterprise additions to the hardware standard in that quarter century. Management and security have been left to individual companies; a BIOS password here, a trusted platform chip there, but nothing that can be specified to apply across the industry in a standard form.

Intel's technologies are well-specified, have a good track record and a measure of independent support. They do things that need to be done, and Intel wants them to be universal. While this is clearly headline news for AMD, it will also stir up the PC makers. Dell has made a good business out of platform stabilisation: get your software image working on a client and you can be sure that the boxes you buy a year from now will run that image as is.

Intel's temptation will be to make the business platform specification good enough to create similar guarantees across manufacturers. In that case Dell can expect much more competition at home and abroad – expect a new concentration on measurable reliability and effective power management for all components.

There's still room for mishap. Intel does not have a perfect track record in defining what's best for the industry. Previous attempts to set the agenda for RAM technologies have badly misfired, and locking Centrino to Intel's own relaxed agenda for wireless has done 802.11a no favours at all.

A crucial aspect of the original PC was that it was open: it was designed around an Intel processor and support chips, but those functions could be and were duplicated by others. That isn't the case for Centrino and won't be for whatever blend of Ts that Intel sells under its own label.

A measure of Intel's own paranoia is called for. Look for areas in the specification which are dependent on one supplier, be it Intel or someone else, and ask what happens if that supplier can't or won't play the game. A platform with one wonky leg can be worse than no platform at all.

Editorial standards