Linux – now Intel stabs Microsoft in the front

Can Wintel survive Linux? The news that Intel, in close cahoots with Netscape, intends to buy a share of the Linux action on Tuesday makes it doubtful, coming as it does on top of a stack of ominous signals from the chip giant. On Tuesday Intel's Sean Malone will share a panel with Netscape's Marc Adreessen and Red Hat CEO Robert Young, speaking to a San Jose audience on 'Linux and the Open Source Business Model of the Future.' Simultaneously, according to various Red Hat-sourced reports, Netscape and Intel are to announce equity stakes in Red Hat, which produces one of the major Linux implementations. Even without the stake the bedfellows and the subject matter would be enough to generate more internal Microsoft memos than the DoJ would know what to do with - but as it is, it is Becoming abundantly clear that the Wintel alliance is crumbling, and that Craig Barrett has a wicked sense of humour (Intel chief claims Microsoft alliance still strong). But why should Intel go for Red Hat Linux? And what's the significance of Netscape's involvement? Effectively, Intel is making a serious attempt to cut its way out of the corner its relationship with Microsoft has blocked it into. In recent months it has become clear that Microsoft has been going out of its way to block Intel moves into software, forcing the company to back away from Java and other kinds of software and drivers that would make it easier for non-Microsoft software to run on Intel. When it comes to the Wintel platform, Microsoft reckons it defines the software, and that the software goes rather further into the hardware than Intel would like to be the case. At the same time Microsoft's definitions of the software have increasingly been putting a brake on Intel's programmes. Microsoft has a narrow view of appliances that doesn't help Intel (hinders it, in that most CE platforms aren't Intel), and the whole of the Microsoft-centric network future is dependent on the much-delayed NT 5.0. By snuggling up to Oracle and the cable companies Intel has shown us it sees whole new markets out there that don't necessarily have much to do with Microsoft, and if Microsoft shouts 'stop!' every time Intel starts to do something about them, Intel could find itself shut out of them. So making the break must have been tempting. Meanwhile, alongside the Wintel Grand Plan Intel has a long history of Unix involvement, which is scarcely surprising considering the number of companies running Intel-based Unix servers. Intel's need for OS support for Merced has also led it into alliances with Sun, HP, SCO and various other Unix vendors, and a couple of weeks ago Intel announced it was supporting the Uniform Driver Interface (UDI) initiative, which is designed to produce an industry standard hardware driver interface for different flavours of Unix. Intel's announcement specifically referred to "growing support from the Linux community and its creator, Linus Torvalds," and said an Intel-based reference design would result in UDI-compliant products by the middle of next year - it now seems highly likely that many of these will be machines running Linux. In the past the constraints associated with staying friends with Microsoft probably haven't burdened Intel excessively. Intel's age-old strategy is to do whatever it takes to sell more and more powerful CPUs, and Microsoft's high-bloat OS iterations did a pretty good job of helping this along. But Intel's needs have changed in the past year, as it's seen the low-cost PC business take off, and as it has focused increasingly on the small business market. The company still wants to sell more power, but it also needs to sell more volume, and in its investigations of small business networking it must increasingly have found itself bumping into Linux. When Intel announced its thin server appliance strategy two weeks ago it could just as well have been announcing its Linux small business strategy instead. Small business wants cost-effective solutions that are simple to install and administer, and Intel wants to produce 'appliance' servers that run operating systems that deliver the functionality needed, no more and no less. The clear implication of this is that Intel wants an OS that can be tailored and then offered on easy terms. Intel also doesn't want an OS that forces users to buy additional per-seat licences. As we said at the time (Intel network scheme means war with Microsoft) this means Intel needs a non-Microsoft OS. Microsoft produces one big server OS which remains over-complicated and overly expensive for small businesses. Microsoft also wants a Client Access Licence (CAL) payment for each and every client connected to the network. So Intel had to work with another vendor. The established Unix vendors may have been unattractive because of their desire to control their own implementations tightly (this is particularly the case with Sun), so as a cheap and flexible general purpose Unix, Linux has obvious attractions for Intel's small business efforts. It has other attractions too - it will run on a 386 with eight megabytes RAM, which is something Intel would have hated two years ago, but is the sort of footprint that could come in handy for Intel's set-top box, NC and appliance efforts. It also runs on ARM, with Corel producing a series of ARM-based Linux machines, so from Intel's point of view could be seen as a unifying OS for its CPUs. And then of course there's the recently-announced Citrix Linux client for integration with NT networks. And Wintel? Intel still needs the revenue from all those Wintel PCs, but there probably isn't a great deal more Microsoft could do in retaliation for Intel buddying-up with the enemy, and when it comes down to it Intel needs the revenue it gets, not the revenue Microsoft does. It might not be putting it too strong to say that Intel, because it was finding its partner in Wintel too burdensome and restrictive, has decided to invent another. Hello Lintel? ® Click for more stories