I have some good friends who stand foursquare behind their Linux servers. Each of them is eager to have Linux adopted as the corporate standard, or at least take over a large share of the operational servers at the companies where they work. About half of them work in large corporations and the other half in a sampling of small- and mid-size companies. Guess which ones are having the most difficulty in selling Linux to their management?

I would think the IT management of large corporations would be the most cautious. Even though Linux is obviously finding support among the large computer companies (e.g., IBM, Dell) it does have a non-commercial history and its image might not appeal to corporate managers. However, at least among the group of people I know (admittedly not a statistically significant sample), enterprise managers do find Linux appealing.

Although something of an oversimplification, the element that seems to favor Linux in the enterprise is the same that works against it in smaller organizations--expertise. All sides understand the leg-up for Linux in its low cost as a piece of software; but nowhere is that considered decisive. In part this is because competing operating systems are also relatively low cost. However, the real cost of an operating system is what it takes to get it up and running and keep it running.

This is the cost where expertise is a decisive factor. If you can afford to hire specialized expertise or, better still, simply have it as a part of normal staffing, then Linux (like Unix) is a not a difficult choice. At this stage in its evolution Linux is not the easiest operating system to install and configure for large-scale operations. Most of my friends brag about installing Linux on a workstation in under an hour; but when pressed about establishing a server farm or some other scheme involving multiple servers and a variety of support equipment, they'll admit that not a few hours of blood, sweat, and tears may be involved.

Linux has some deficiencies in the larger IT configurations that are just beginning to be addressed. Support for multiple processors and availability of some kinds of peripheral hardware may be an issue. Most of my friends in the big companies don't see Linux as necessarily appropriate for the corporate computational heavy lifting. Yet, unlike Unix, with its hardware champions such as Sun and HP, Linux doesn't carry the historical burden of being a mainframe or minicomputer system. Its short history binds it rather nicely to the Internet and its use in server farm configurations is growing by leaps and bounds.

Here too the key to Linux seems to be the level of expertise the corporation can bring to its use and as a corollary the degree to which its people can stay plugged into the changing and upgraded Linux code. My friends say they spend a fair amount of time communicating with other Linux aficionados and in a few cases even working on the Linux code itself (on company time I might add).

My friends in smaller organizations who support Linux feel more isolated. They frequently don't have enough time to stay on top of developments, much less participate in the evolution of Linux. They have a tough time convincing management to get them the time or the support. The shortage of expertise--and its high cost--is making the spread of Linux to smaller organizations more difficult. Not every company can find or afford a Linux guru; yet a guru or something like one is what Linux needs. Even in relatively trivial configurations (more than one server), successful installation and maintenance of a Linux-based system requires a good deal of knowledge and experience. This doesn't come cheap, and most managers know it.

While the availability of Linux-savvy workers is improving and more enterprise level companies are beginning to provide support and outsourcing services for Linux; it seems the small- to-midsize corporations have trouble getting beyond the cost and difficulty of implementing Linux servers in number. They've heard about Linux. They know Linux has many virtues. But it seems like yet again, it's the people issues that are the crucial factor.

Nelson King is editor at large for ComputerUser.com. Also read his monthly Pursuits column in ComputerUser magazine.