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

Some years back, one of my colleagues wrote an article in which he committed a mild blasphemy against the Christian faith. That week, we got more letters for publication than for the rest of the entire year.

So it is with a sense of trepidation that I now intend to blaspheme against another religion: open source computing. It is my contention that the benefits of open source software are greatly exaggerated, and that the open source 'community' is and will continue to be far less important to the business world than is widely perceived. In fact, were it not for the commercial self-interest and the clever marketing of suppliers, all of this would already be widely held to be self-evident.

There are many reasons why the open source model for producing software is so highly regarded, but lets cut to the chase and identify the main one: a deep and widely held suspicion of big business, and in particular, Microsoft. Nine times out of ten, that translates into the simplistic debate of the Windows v Linux operating systems, in which the latter is always the good guy.

Forrester Research highlighted this in a report on security in April. After collecting a year's worth of data, the analyst group concluded that both Windows and Linux can be deployed equally securely. But, it adds, "Microsoft gets a bad rap for security, while many believe that Linux is relatively secure". In fact, on two or three occasions, reports critical of Linux or open source have been lambasted, partly because Microsoft helped to fund them.

But to focus just on Windows v Linux is too narrow. There are several good reasons why, while individual products such as Linux might be viable and important, the open source production model itself is not.

Let's take, first, the simple economics. Over the past two decades, software as a percentage of a total IT solution has risen as the proportion made up by hardware has fallen. That has made the case for free software much stonger; but in the coming years, the price of software will increasingly be embedded into an overall managed service or solution. This can be seen everywhere - in Microsoft's Software Assurance scheme, in the grid computing model, or in application service provider offerings such as Salesforce.com. From the user's point of view, then, the cost of the components is of little consequence; what is important is the total cost. This can be a tricky calculation: Linux solutions are not always cheaper - and where they are, it can often be attributed to many factors.

Another misunderstanding is that most of the big IT suppliers - such as IBM, HP, Oracle, SAP and Sun - are in favour of open source software. But they aren't: rather, they are in favour of low cost components. Linux was the perfect candidate for open sourcing - first, because suppliers were burdened with developing and supporting too many versions of Unix, none of which offered a real competitive advantage; and second, because Microsoft was threatening to gain too much market share. To measure the real depth of support for open source, ask Oracle or IBM if they are in favour of the MySQL database, or BEA if it is favour of the JBoss open source application server.

Another myth is that there is a vibrant community of open software suppliers. There isn't: there are a handful of companies, most of them small, barely profitable and strategically challenged. Red Hat, the king of the specialists, has been shunned by some open source purists because its model has changed - its premium software now comes with a mandatory service charge.

And what of the army of individuals writing the code? Yes, they exist and there are a lot of them - but as the legal squabbles between SCO, IBM, Red Hat and others show, a big chunk of Linux code now comes from big suppliers. That production model is not so much the bazaar (to cite the tortured open source analogy) as a consortium of cathedrals pooling their resources.

But users love it, don't they? That is not clear, either. At a recent IT directors debate, all those present agreed that that they would only use Linux where "there is a supplier to kick if things go wrong". And none would trust their critical infrastructure to someone they are not paying.

The upshot of all this is that there are two types of open source in the world. Linux, backed by IBM and HP, as a proprietary software replacement and increasingly used as a Microsoft or Unix alternative; and the rest, which is little used and has little credibility.

All this may well change - but not in the way that open source theorists such as Linus Torvalds or Eric Raymond envisioned. The future of corporate software is service oriented, in which components and services are plugged into a managed infrastructure. The big suppliers such as Microsoft, IBM, HP and Oracle will take the bulk of the customer's money, and they are not thinking of open source as a cash cow. And if some of those services are open source, they won't mind. But most of them won't be; most of the suppliers, whether individuals or companies, will prefer to be paid.