The Linux operating system (OS) has just taken what is being touted as a giant leap forward, with the eagerly anticipated, but well overdue, release of version 2.4 of its kernel.

And initial excitement has been high within the Linux community.

Everyone, ranging from the techie anoraks who devote their free time to helping develop the open source OS, to the hard-nosed commercial distributors hoping to take it to the heart of the enterprise, has greeted the new kernel with joy and relief, pending a thorough road test.

But less certain is the response of the corporate user base that the kernel has been so carefully groomed for. Nor is it entirely clear exactly when usable 2.4-based products will be available for them to deploy.

All eyes on Linus

And the sheer length of time the new release has taken to gestate has added weight to the significance of such questions. As a result, all eyes are now watching and much is at stake.

Linux creator Linus Torvalds initially said in June 1999 that Linux 2.4 would be ready for developer consumption by autumn 2000.

In May 2000 he acknowledged, however, that 2.4 was likely to be delayed until October 2000 because developers were attempting to include more new, high-end functionality into the final release.

But last October, at LinuxWorld in Frankfurt, Torvalds admitted that Linux 2.4 wouldn't be launched until December at the earliest. As the press started work on 'Linux late again' stories, he decided on 4 January to cut the apron strings.

In a round robin email to the Linux community, Torvalds said he had "decided that enough is enough, and that things don't get better from having the same people test it over and over again".

So in short, 2.4 is now available and the kernel appears to contain numerous improvements.

What's new in 2.4

For the desktop, the most obvious benefit is support for devices such asprinters, digital cameras, scanners, keyboards, mice, network cards, modems, zip drives and products that plug into a universal serial-bus port.

For those concerned with network infrastructure, the hottest changes include enabling Linux to take better advantage of systems with up to 16 processors such as IA-64, S/390 and SuperH handheld-device chips.

And this will be the single biggest factor in driving the OS into high end server space. The 2.4 kernel, should, in theory anyway, improve Linux's performance in running data centre applications such as those already on mainframes, Unix servers, and high-end Windows 2000 boxes.

It also supports bigger physical memory on Intel-based servers, which is necessary for it to run many high-end databases and server applications.

Input from distributors

Before any of these goodies will be available to the broad mass of users, however, the various Linux distributors will need to add their input first, packaging up raw 2.4 code to create their own Linux implementations.

This, say many analysts, is a truer test of Linux's future than the 4 January announcement. Companies such as Red Hat, TurboLinux and Caldera have endured a tense few months waiting for the new kernel, since, unlike Torvalds, they have investors, analysts and real live customers to please.

No doubt if Torvalds had decided to delay the new kernel by another six months, we would have seen the first of these firms go to the wall, taking with them Linux's best chance of making the enterprise grade.

Even so, the distributors can't afford to rush things too much. TurboLinux, for example, says it plans to make sure that its forthcoming 2.4-based release can fully support mission-critical applications, and it claims it will not considerreleasing anything that falls short of that any quicker.

Red Hat has likewise been planning to make the 2.4 kernel the heart of its next release, codenamed Florence, which is due out in the first quarter. Caldera Systems, on the other hand, intends to upgrade both its eDesktop and eServer Linux releases to support the 2.4 kernel in the second quarter.

So both are now up against it. All of these plans will be up in the air if the 2.4 kernel is too flawed for commercial use, but nobody will know anything for sure for a month or two yet.

Patches, propaganda and hype

This means that, for a while, the Linux community is likely to find itself in a similar position to Microsoft following the launch of Windows 2000. Here most corporate users have been wary of moving established applications and data to a major new revision of the OS, preferring to wait for patches and fixes.

And because the majority of existing Linux 2.2 packages are backward-compatible with Linux 2.4, there is really no reason for corporate network managers to rush into any decisions. Time is on their side.

As George Weiss, Gartner's research director for server and operating systems, said: "Users don't run mission critical applications on .0 releases. It will be at least six to nine months, maybe even a year, before we can judge whether it lives up to its reputation. To say it will mean the death of Risc Unix is propaganda and hype."