InformationWeek

November 15, 1999

TechView:
Linux Has A Core Role In IT, Not Business

Gregory Smith ith a final release date for Windows 2000 set at last, I ought to once again stray into the Operating System Jihad. Clearly, Windows 2000 and Linux have dominated discussions of operating systems for most of the year. I dare say Linux is tried-and-true technology, and there's little contention that Windows 2000 isn't really a product yet. But both the hoopla about and the relative maturity of these two platforms seem to lead people to the wrong conclusions.

I can safely say that I love Linux--at least in the abstract. It is, with good reason, this year's darling of the open-source community. The anarchist in me loves open source--and so does the propeller that occasionally pops out of my head. An awful lot of the truly fundamental advances in IT infrastructure have come from open source and whatever we called the important parts of the open-source idea before it had a name: Unix, C, Sendmail, and Mosaic, to name just a few open revolutions. Perhaps the most potent effect of freely

distributed, robust, open software is that it enables other application developers to assume the presence of some critical piece of infrastructure. Whether it's a multitasking operating system, a nearly universal programming language, E-mail for everyone, or a front end from some new markup language, open-source software lets enterprising developers create applications that do something marvelous with a new brick in the IT infrastructure.

Of course, proprietary software that's free can have the same effect. Look at how Adobe's Acrobat Reader and Netscape's Navigator improved rich information exchange throughout the computing world. There's no doubt that these were potent forces of change in the widely connected world of computing.

As for my propeller head: The more source code I have, the happier I am. I can "borrow" good ideas that I find in the code. If I try hard enough, I can even fix the bugs that are invariably a part of every product I've ever used. Imagine that. No more calls to Microsoft to find out that they know about my problem and, if I send them $49 next year, they'll have an upgrade ready that will solve my problem. A lot of good that will do helping me get a system running now.

But it's just this scenario in which the real importance of open source starts to unravel. If you're running one of those very rare IT shops with almost no turnover among your stable of very gifted programmers who are intimately familiar with the inner workings of your open-source platforms, the self-supporting possibilities offered by open-source software puts you in an enviable position. Ditto for the potential power of applications built on the known behavior of the innards of an operating system. For the rest of us, however, anchoring our IT efforts to a cadre of programmers with an intimate knowledge of and the ability to recompile entire operating systems belongs in the catalog of worst business practices.

Competition, more than open software, is what really drives innovation and open standards that enable customer choice. It's hard not to like open software, but exactly how does that kind of openness further your strategic goals? Linux is not the revolutionary product that Domino was (when it was called Notes, of course), and there's nothing open about that platform. That you'll be able to run Domino on Linux instead of NT in a matter of weeks might have some appeal, but it need not be a strategic choice.

Linux has a place in IT, but it's a tactical one. If you need to set up an intranet quickly or prototype a Web site, Linux lets you install over the Internet quickly and will run well with surprisingly modest hardware. While Web and E-mail services are the most common applications for Linux servers, there are certainly other applications that make sense for Linux. For corporate infrastructure, however, Linux looks a lot scarier. Is Linux reliable? There are scads of Internet service providers who run tons of Linux servers all day every day without any problems. But most of these boxes are only loosely managed and scaled in ways that are impractical for enterprise applications--typically by adding more essentially independent servers.

But at a strategic level, Linux just doesn't play. Initiatives that will transform your business include: better-integrating enterprise resources, consolidating management with a unified directory, deploying systems that are designed to scale well and run reliably, improving workflow and knowledge management throughout your supply chain, and cementing relationships with your customers. Whether Linux serves Apache pages faster than some NT system on a particular piece of hardware is an interesting tactical debate, and I'm sure Microsoft is worried about losing some business on NT licenses.

As much as I love Linux, at a strategic level, it's still a distraction and unlikely to be much more in the coming years. Deploy Linux where it's known to work well if you've got the talent to support it--but as a core platform for your strategic initiatives, it's just not there.

Back to the Columnist page
Send Us Your Feedback
Top of the Page