Linux rises to top dog in servers

LAST week's LinuxWorld talkfest in San Francisco, which brought together 10,000 vendors and users of the open source operating system, offered yet another indicator of its growing popularity across enterprise IT groups and among developers.

The steady march of Linux into ever more central areas of enterprise IT has been visible in the US for about seven or eight years, and has accelerated in the past two.

Once consigned to "non-essential" functions such as serving web pages, Linux is increasingly the basis for software stacks that perform critical data centre tasks.

This trend is likely to continue, and is a remarkable success story for software once ridiculed as unfit for corporate use, but the focus at LinuxWorld was not the story to date, as much as the future - where Linux and the open source apps that often sit atop it may be headed next. The rise of Linux in enterprises has so far come largely at the expense of older, proprietary flavours of Unix such as Hewlett-Packard/UX, AIX and Solaris, rather than Windows, although this is changing and the more direct open source challenge to Microsoft's market share is growing.

Having largely lost the battle to protect their own versions of Unix (at least for all but the most demanding applications) HP, IBM and more recently (and reluctantly) Sun have become the strongest vendor supporters of enterprise Linux, since this allows them to keep selling boxes.

Of the three, the company that has committed most heavily to Linux is IBM, and some of last week's more thoughtful comments on what might be the key near-term directions for the industry that has sprung up around the operating system came from Bob Sutor, IBM open source vice-president.

Sutor's view on the main immediate drivers of continuing growth were not particularly startling: the economic and environmental need for greater energy efficiency in computing, server consolidation, virtualisation, and the spread of Linux from x86 to other processor families are all fairly clear trends.

Sutor was surprisingly upbeat on the prospects for Linux-based systems among small and medium IT users, where it has so far failed to make as much headway as in larger environments, and Microsoft remains king.

The free-wheeling, organic nature of open source software development and the lack of any co-ordinating authority means that to this point, there have been fewer of the seamlessly integrated, simple, locked-down systems that smaller IT shops increasingly prefer to "rolling their own".

Linux may yet play a larger role in serving these users as IT functionality moves to the cloud, where it is strongly positioned, and as hardware or system vendors work harder on piecing together products that eliminate the complexity of integration that deters many smaller users.

More pessimistic was Sutor's view on the prospects for Linux and open source systems in the vertically focused business applications space.

Here, he predicted, a few industries could embrace open source top-to-bottom systems if more effective versions of these emerged, but users in many others would stick with proprietary industry standard vendors and answers at the app level, albeit with broader acceptance of Linux as a supported underlying operating system.

One additional frontier inside the enterprise often cited by open source fans as an opportunity is the desktop, but it is clear that this remains challenging terrain despite a recent uptick in Linux installations on new machines and buzz about friendlier desktop distributions such as Ubuntu.

Again, it is most likely Linux and open source will win share in this part of computing via the migration of many of the functions performed by today's desktops into the cloud.

The more traditional (and traditionalist) dream of open source desktops as polished, user-friendly and widely accepted as Windows or Mac OS continues to recede into the future.

That doesn't detract from the remarkable rise of Linux elsewhere, and most spectacularly in enterprise IT, where its proliferation shows no sign, as yet, of slowing down.