Pondering on the future of the operating system has become quite fashionable, prompted largely it seems by the growth of virtualisation. Trouble is, everyone seems to be looking forwards -- not backwards.

When it comes to operating systems, the prevailing wind -- to paraphrase Claude Rains in the movie Casablanca -- hails from somewhere other than Redmond. Or at least, that's what we're led to believe.

The OS is going away, they say

The forces ranged against Microsoft now appear staggeringly huge. From being almost the only fish in the pond, Microsoft is now almost surrounded, with VMware and its virtualisation technology as cheerleader.

The whole open source and Linux movements remain hugely opposed to Microsoft -- while, incidentally, they attempt to match the OS giant's products with software distributions that are even bigger: Red Hat's Fedora Core OS comes on five CDs while Windows XP manages with one, for example. Admittedly, there are many, many more applications on those five CDs...

But even Linux folk are not immune to the siren call of the thinner OS -- a call that's been heard many times before over the last 15 years. One key argument runs that the growth of virtualisation means that the OS has less to do. With hypervisors metering out hardware resources to virtual machines -- CPU, memory, disk, I/O and so on -- the role of the OS is seriously diminished.

And hardware is no longer important, they say. The extent to which you don't care about the hardware -- which has become seriously commoditised -- is demonstrated by the fact that you can now run Windows applications on Apple Mac hardware using virtualisation systems such as Parallels Desktop for Mac.

In a fair world, it also wouldn't be long before running Apple OS X on generic x86 hardware became commonplace. But for that to happen, Apple needs to give up its obsession with proprietary hardware, since it's written into the licence agreement of OS X that thou shalt not run it in a virtual machine, or on anything other than an Apple-labelled box. Although we've been waiting for that move for at least 15 years so I'm not holding my breath.

Alongside that is the argument that Java does something similar to the hypervisor but a tad higher up the stack. Once a Java VM is up and running, the need for an OS goes away. This argument works only for server-based enterprise applications of course -- desktops don't enter into this one. But a JVM does do more than a hypervisor and has been described as being closer to an OS than a virtualisation hypervisor.

When it comes to desktops, other devices are getting smaller and taking over desktop functionality. Think about smartphones, Blackberrys and PDAs, and suddenly, the dull old desktop with its humungous OS starts to look obsolete.

Gartner has hammered this notion home by predicting that Vista will be Microsoft's last version of Windows -- something Microsoft boss Steve Ballmer denies. Instead, says Gartner, it'll be a modular OS and sold on-demand by subscription as a service.

Will Microsoft's core business disappear?

Everyone sees Microsoft's core business disappearing -- apart from Microsoft -- and a fair few hardware vendors.

Well, maybe.

Microsoft has already trumped this prediction by announcing that there will be a follow-up to Vista, probably codenamed Vienna. It's due for release in 2009 but seasoned Microsoft watchers will understand our guesstimate that 2011 might be more likely.

And Gartner's got it wrong, says Microsoft, arguing that people don't want to run software over the Internet. And, to a large extent, Microsoft is probably right, since there isn't a groundswell of people demanding to run software over the Internet. If users want something other than Office because of its high price, there are now free alternatives. And you can't beat the performance of code running locally on your own CPU, using data that hammers into the machine from local storage.

And in fact history tends to confirm this position. The coming of the thin client, the future of software as a service and a number of similar initiatives have been trumpeted on a number of occasions but none has displaced the OS or the fat client in mainstream computing. People always want more.

A lot of that has to do with the way that OSes continue to offers users more -- in other words, they keep getting richer. You can tell users like that because they keep coming back for more. The same thing happens with mobile phones. Despite the urgings of some for basic phones: the ones that sell are those offering more features -- fatter clients, if you like.

By looking back on that unbroken trend from the past, we can infer the future. There's no reason to suppose that the OS as a concept is going to change radically -- even if it might get fatter.

What's more, since Redmond has made a number of calculated guesses over the 30-plus years of its existence and got most of them right, you'd be wrong not to take account of the company's views of the future of the OS -- especially since it's in a position to determine much of that future.

Follow the money

It's just in the interests of a lot of people who aren't selling OSes to suggest that the world's going to change tomorrow. None of those people is called Microsoft, most of them have something else to sell, and they can't all be right. Incidentally, those few others who sell alternative OSes aren't arguing for the OS's disappearance either.

And despite the suggestions that hypervisors and JVMs will overtake the OS, past experience suggests they won't. Like most new technologies, at best they'll sit alongside existing ones for quite some time.

Granted, the past does is not a reliable guide to the future. But if the future truly is going to be hugely different, you have to ask what's going to change. And when you do, it becomes difficult to imagine that the need for local intelligence and software to manage hardware resources will go anywhere, anytime soon.