OK, all of you fellow Old Farts who understand the title skip the next three paragraphs. For you newbs, here's some back story:
Once upon a time, there was this box called an IBM PC AT, which featured a kickin' 80286 processor, a substantial upgrade to the 8088 in the original PC. The two machines shared a common bus architecture, which allowed the use of plug-in cards to expand functionality. Back then you needed an expansion to do just about anything, including run the display. Trouble was, IBM had openly published the specs for that bus, as well as the rest of the original PC, and a lot of 3rd parties had started punching out expansion cards to compete with mother IBM. Even worse, there were 'cloners' who were copying the entire machine, and selling against Big Blue. This was not Good For Business and something needed to change.
When the 80286 generation rolled over to 80386, IBM tried to make that change. They ripped the standard PC bus out of the design, and substituted something called the Micro Channel Architecture (MCA), which was of course proprietary and patented by IBM. That would be the end of the license-free 3rd party add-ons, and of truly 'IBM compatible' PC clones.
In spite of some claimed architectural benefits of the MCA, the business intent was abundantly clear to both vendors and customers, and the market rejected it. The 'cloners' built 80386 boxes around the PC bus architecture that ran the PC-DOS of the day without modification. IBM's MCA based PCs entered the market late and ran a poor second to the cloners' machines. IBM was eventually forced to withdraw MCA from the market. Big Blue suffered an enormous erosion of PC market share and permanently lost strategic control of the PC platform to Microsoft, which happily supplied software to the cloners.
I'm not suggesting there is a complete parallel to Microsoft's situation with Vista, but there are a lot of similarities: A product introduced for strategic advantage at least as much as customer benefit. A workable alternative from the same vendor that remains in the market and still has wide support. Platform level competitors emerging as an alternative solution. Let me count the ways:
After the lengthy gestation of 'Longhorn', the eventual Vista release has had to deal with a market well-saturated with Windows XP, including an enormous number of pirated copies. To deal with the piracy situation, and likely as a hook to pursue its long-term goal of moving customers to an obligatory subscription model, MSFT put in its 'kill switch' to remotely turn off copies analyzed as bogus. Like all copy protection and DRM schemes, it proved to be fallible, and customer ire forced Microsoft to drop the feature.
Vendors are forced into that type of backdown when the product fails to achieve 'must have' status. Like the Micro Channel, Vista appears to be a product the market can live without. Many industry users appear happy to have XP live forever. Of course, Microsoft can try to force XP out of the market, by cutting off support and upgrades, particularly security fixes [snark]. But that could easily backfire in removing its ability to control a force that has its own momentum, particularly in the gray to black market stronghold of China. Analysts are seriously suggesting Microsoft should quietly back down on Vista - a la ME - and instead upgrade XP. Meanwhile, the first bug fix release for Vista seems to be causing more problems.
It is possible to write off some amount of this to attention getting behavior of bloggers and IDG columnists. Microsoft has faced such problems before - 3.1 to 95, 95 to XP - and has eventually outlasted the critics and picked up the late and grumpy adopters. What might be different this time? Increasing strategic competition.
I'm not suggesting that someone is going to reverse engineer and open source or sell XP against Microsoft, though one can imagine a pirate XP fleet living on in China for some time. But Microsoft has always worked hard to be sure that its operating systems are well-nigh obligatory for any box that has an Intel-architecture chip. Those who have watched the industry for some time have observed Intel regularly try to expose an API independent of Microsoft, and just as regularly be slapped down by MSFT using its market leverage. (Dig back into the histories of sound and telephony APIs, for instance). Microsoft has worked hard to be sure no architectural layer intervened between its operating system and the iron.
Intel is now getting its wish, as Microsoft loses control of that lock-in. Some part of this story is old news. Linux has been an equal competitor in the server market for some time, particularly in Web and eCommerce shops. But it's another force coming out of the data center that should be frightening Microsoft: virtualization.
Venture investors got excited about this market with the stellar VMware IPO in mid-2007. From about 12 venture funded, virtualization related companies at the end of 2006, the field ballooned (bubbled?) to over 50 by the end of 2007. Unsurprisingly for a spinoff from data center giant EMC, VMware was largely about addressing the need of the server side to streamline management and reduce hardware costs by improving flexibility in provisioning operating system platforms and applications capacity. But there was also potential beyond the server farm.
I had my eyes opened in early 2007, as I watched a developer use VMware on a PC to run Linux and Windows side-by-side, even setting up throw-away operating system images for test purposes (simulating a distributed system on one box, for instance). The Apple world had begun seeing this potential even earlier, with Parallels for Mac providing the ability to turn a Mac laptop into the world's fastest Windows box, while leaving OS X running as well.
I wasn't the only one to notice, and there are an increasing number of virtualization ventures aiming at client / desktop usage, as well as more selling in that direction from established virtualization vendors. In addition to the develop and test uses I observed, virtualization on the end user PC addresses a number of the configuration, patch maintenance and security upgrade needs that are ongoing headaches for IT managers, and create the demand for existing vendors such as BigFix. For instance, a preconfigured operating system image can be loaded and launched on client from a central site, and that image terminated if it becomes corrupted or infested with malware.
Why is this a threat to Microsoft? First, because a fundamental component of a virtualization system, the hypervisor that effectively acts as a smart boot-loader, can run directly on the PC hardware ('Type 1' in the linked Wikipedia article). A hosted operating system, such as Windows, runs above that layer and is within its control. The hypervisor is therefore exactly the intervening architectural layer between Windows and hardware that Microsoft has long feared. By presenting an alternative way to expand functionality, rather than needing to host it within Windows, it breaks a strategic element of platform control.
That may sound a little abstract. Here's the practical possibility: Keep Windows XP on current client machines. Add a hypervisor layer beneath it. Use the virtualization platform to manage configuration and security to justify the current expense. Now run experiments with alternative operating systems, such as Linux configurations, alive and side-by-side with XPs. Introduce Vista sparingly, using virtualization to manage a pool of licenses across the enterprise, again layered over the hypervisor. See which options work out the best for users and managers, meanwhile avoiding upgrade expenses.
It's the possibility of this option being taken seriously in large enterprise customers that could make the brouhaha about Vista's problems into more than the carping of pundits and late adopters. Like IBM, Microsoft could find the down side of trying to push the market where it doesn't want to go, with its own XP playing the part of the old PC bus in a shifting platform world.