At the end of the year, it's traditional to look back at the big stories of the year, and ask which will prove significant for the future.
Clearly, Microsoft's launch of Vista is a big event in the industry, not least because the livelihoods of hundreds of hardware vendors rest on its success. While that remains true to an extent, it's arguable whether Microsoft's grip on the desktop OS market is anything like as rigid as it was when XP was launched, those six long years ago.
From an enterprise perspective, Vista doesn't register. Most large companies won't be considering desktop upgrades for some time, unless their refresh cycles just happen to fall in 2007. Even then, they might well consider delaying the expenditure -- never that hard a decision -- until the Vista hype has settled. IT managers we've spoken to reckon that since they've just finished upgrading the desktop estate to XP, a move to Vista is not imminent.
And then the danger for Microsoft is that an upgrade cycle creates opportunities for rivals. And today, there are two rivals.
You'd have to have been on Mars to miss the upwelling of interest in Linux. As a desktop OS, it's become more usable which makes it a better bet. There are far more reference sites out there too, and the problem of unavailability of technical skills is rapidly melting in the furnace of interest. Those skills are far more widespread than five years ago, making them cheap and easy to come by.
The other rival is virtualisation. It's a rapidly maturing technology, and the number of vendors interested in and selling desktop virtualisation systems is growing. This implies that the much-touted move towards thin clients could be driven not by terminal services technology but by virtualisation. This technique offers the advantages of keeping computing power at the desktop while retaining central control. It's also another cost-justification for those power-hungry data centres where the desktop VMs will be hosted. Desktop virtualisation is not mainstream today, but 2007 could see this market becoming significant.
Meanwhile, we've also had early sniffs of how the virtualisation software market could start to develop from a one-horse race into a multi-player field. The imminence of Microsoft's entry will concentrate minds and force development, as the entry of players such as Virtual Iron and others could well start to see VMware's almost impregnable position start to become eroded. Expect more price falls, and more paid-for products to become free.
And while we're talking OSes, Longhorn -- the next version of Microsoft Windows Server -- is due by the end of 2007. But as we all know, it's best not to hold your breath. An OS is an amazingly complex piece of software and slippage is more likely than not.
The other huge trend has been both the acceptance by the industry of the need to cut power usage and the arrival of robust products implementing that philosophy. From data centre equipment vendors to CPU designers, from facilities managers to those running electricity grids, the need to cut power usage is now top of mind. With that much concentration of brain-power on a single topic, expect to see some innovative solutions in 2007.
One effect on the hardware market of this trend has been a shift to fewer, bigger servers, much to the delight of the likes of HP and IBM. Their big Unix servers now become the ideal types of platform on which to run a virtualised infrastructure -- although the rather unseemly name-calling cat-fight in which both vendors are currently indulging makes the whole business seem a little sordid. Mine's better than yours, they cry.
What's not succeeded in 2006 is the notion of utility computing. In 2005, Sun promised its customers that they could buy compute power for a dollar per unit but, with few takers, the company had reluctantly to give up the notion in 2006. Customers don't get grid, complained Sun's boss Jonathan Schwartz.
2006 was also the year when AMD consolidated its entry into the market as a builder of enterprise-level products. Of course, hardware vendors have been designing servers around its industry-leading, power-efficient Opteron chips since its launch in 2003. But this year, after watching its rivals chew up its Intel-only market share as enterprise customers demanded less power-hungry products, Dell decided that enough was enough. It concluded its Intel-only policy and quietly slipped out AMD-based servers.
And Intel itself managed around mid-year to launch -- to huge sighs of relief from its Santa Clara HQ -- its Opteron competitor, which fulfils the promises that Intel made on its behalf. It was a long wait.
Other chip technologies have been less fortunate. This year saw the announcement of the end of support from HP for Alpha. It'll support AlphaServer until 2011, but availability became uncertain as the company decided in October to make no more of them. Slightly more optimistically, Intel's high-end Itanium processor continued along its bumpy but flat-lined market share curve. Those who use it swear by it, while most others remain unmoved. It's a tour de force as a technology, but Itanium seems to be answering a question few have asked -- or at least, that most already have an answer for. Despite that, IDC in February saw a bright future for Itanium. We'll wait and see.
The open source movement took one step forward with the growth in the availability of open source hardware -- even Sun jumped on this bandwagon. But with Oracle's decision both to ship Red Hat Linux with its eponymous database, and develop it along a separate thread, the fear of forking or worse reared its head. One step back, then. The impact of this decision on Red Hat has also yet to be seen -- and the two are now head-to-head in the fiercely contested middleware market, Red Hat with its 2006-acquired JBoss division, Oracle with its long-established database.
The rest of the open source community took heart from the likes of VMware opening up its disk file format, allowing developers to write applications for its products -- and inter alia putting more clear blue water between it and Microsoft.
Clearly an interesting year and 2007 will see continuation of those trends, with the fight to extract more computing power from within a rigidly confined power envelope becoming an over-arching limitation.
Happy New Year!