Information Age: News, analysis & insight for IT & business leaders

When Microsoft's Steve Ballmer interrupted a skiing trip in the Spring of 2003, and flew to Munich to plead with a customer, many observers thought it was a symbolic moment (Information Age, July 2003). Ballmer is notoriously, even belligerently upbeat, but here he was, playing the role of the supplicant, vainly trying to dissuade the burghers of Munich City Council from dumping Windows in favour of Linux.

Munich may be a big city, but it is no New York, and still less is it General Motors, or General Electric. In the league tables of IT spenders, it would probably not make the global top 5,000. Nor is it normally associated with being a technology trend-setter. Even so, its decision - in favour of local supplier SuSE Linux - sparked headlines and speculation around the world that Microsoft had met its nemesis. Linux, the articles read, is free - and how can anyone compete against something that is free?

Shortly after that, according to Microsoft insiders, the company began to examine where things were going wrong. Citing the mantra "it's only free if your time is worth nothing", Microsoft executives argued in a series of meetings that Linux is definitely not free, because it costs more to use than Windows. The trouble was, Microsoft did not have much data to prove its case, and even where it did, many customers did not believe it.

The answer: a new strategy that included a public relations assault and the sponsoring of a variety of reports and research from recognised, 'independent' sources. The goal was to show that when it comes to solving practical problems, a Windows environment is almost always cheaper and more effective than Linux - even when the latter is free.

The fruits of this initiative have been seen in the past three months. And, almost inevitably, the results have been deeply controversial.

Pressure and bias

The first report, 'Windows 2000 v. Linux in enterprise computing' was sponsored by Microsoft but carried out by IDC, the well-known and reputable research company. IDC's research focused on the cost of running enterprise servers - and unsurprisingly, it found that staff account for nearly two thirds of the costs. And when the other costs (such as extra software) have been added in, the researchers found that the fact that the software licence is free was soon diluted. Windows, the report said, is cheaper for security, for printing, for file serving and networking.

So far, so good for Microsoft and Windows. But then the trouble started: Business Week accused IDC of bias and succumbing to pressure from Microsoft in its choice of workloads. IDC denied this was the case, but by then, the impact of the report had already been diluted.

Microsoft also commissioned a second report from Forrester Research. This focused on development, and in particular on new portal style applications using Microsoft's .Net development and deployment environment. It found that .Net (and Windows) is almost 40% cheaper in large and mid-sized organisations than when using a combination of alternative tools based on Java and Linux.

With the IDC controversy still fresh, however, many observers began to suspect Microsoft was stacking the odds. Coverage of the findings focused on the small sample (12 customers) and the fact that Microsoft had again funded the research.

But, fortunately for Microsoft, there was more to come: Gartner Group released its own report, analysing whether the costs of migrating from Windows on the desktop would outweigh the benefits of staying with Microsoft. And this time, Microsoft did not pay for the research.

The findings: moving desktop systems from Windows to Linux only saves money in certain specific situations and is unlikely to be economic for users of the more recent versions of Windows, such as Windows 2000 and Windows XP.

Goal-less draw?

What do all these papers tell the buyer about the relative economic merits of the various systems? The first and main point is that there are few definitive answers: in some cases, Linux saves money; in others, Windows (and .Net) save money. Moreover, even where the costs do favour one or the other, there is, as yet, little reliable, independent supporting evidence of this. Those calculations need to be made using the customer's own data; only in some cases - perhaps such as Munich - will the findings clearly favour one choice.

This, however, is not a goal-less draw, but a victory for Microsoft. As Forrester Research has discovered, most businesses choose Linux to save money. But if the case is unproven, or the evidence of this is ambiguous, then fewer customers will be inclined to take any unnecessary risks or investigate making a switch away from the incumbent.

That does not mean that Microsoft will have a trouble free ride against Linux - there are still some big battles to come. But since Munich, there have no big defections, and reports of at least one huge big Linux win over Microsoft (Ford Motors) proved to be untrue.

Indeed, much of the evidence is suggesting that Linux will, initially at least, find most of its customers from users of proprietary Unix products, such as Sun's Solaris. Here, the cost comparisons are far cleaner and clearer.

If Linux - or indeed the open source software movement - is to be the undoing of Microsoft, it isn't going to happen very fast - if at all.