The Linux Server's Place in the Enterprise

By S. Thomas Adelstein, CPA
CFO, Bynari Systems
Dallas, TX

At the Midway Point and Beginning to Scale

With a checklist from D. H. Brown*, the Linux Community has an Enterprise Blueprint. Those closest to the Open Source Community say, "It's not a matter of if, it's a matter of when Linux will reach all the functional requirements of the enterprise." It may seem ironical but IBM AIX 4.3.2, Compaq Tru64 UNIX 4.0F and SUN Solaris have provided the benchmarks for the next leg of the Linux development effort.

At the monthly UNIX Users Group of Dallas, Jon "mad dog" Hall presented about two hundred attentive listeners with a sobering talk about Linux. Jon Hall is the President of Linux International and works for Compaq. He's been one of the major advocates of Linux since the now famous operating system went public in 1994. The following commentary came about after his speech and a thorough reading of the D. H. Brown report.

I've noticed that people often put Linux in the same category with the UNIX operating systems (O/S). A major difference exists because Linux is not UNIX. Linux developed from a few thousand lines of code into a remarkable operating system. Analysts see Linux as worthy for low-end to mid-sized server applications. In many cases, IT departments need Linux to provide functions not available in other systems especially with Internet protocols. I'm one of those people.

With all the attention focused on Linux, few people realize Linus Torvalds and his team started developing a system they and others could afford and use on the Intel 386 processor. By chance people began putting the source code and compiled applications on inexpensive media. Also by chance, IT professionals deployed it in their organizations, first experimentally now, seriously. The original developers did not necessarily intend to develop a commercially competitive product. The product actually found its own way into the market.

Now, one of the most prestigious analysts in the industry has written an analysis of the system. D. H. Brown says in their report, " In key application areas that employ open protocols or well-characterized closed protocols, such as low-end or mid-range Web serving, e-mail routing, network printing, and file serving, Linux can provide a solution that once properly configured, is both stable and performs adequately for at least moderate workloads. Further, Linux's "good enough" capabilities come at minimal cost and do not incur significant vendor lock-in." (From "Linux: How Good Is IT", Port Chester, NY, 1999)

Linux is not UNIX and it's not a UNIX clone. The Linux design has always allowed people to add the functionality to it. The Open Source nature of the software allows it to become an excellent enterprise system. The developers did not target the enterprise during its creation of the system. That's the serendipity of Linux - the idea of sharing and working together and allowing others to continue that work. Somehow that doesn't fit a model of the world with poverty as the context. Frankly, it fits a model where prosperity and vision form the context.

The Linux development community shares few of the limitations imposed by UNIX. D. H. Brown Associates, Inc. compared Linux Red Hat and Caldera Open Linux to conventional UNIX and Microsoft NT Server. Their assessment provided Linus Torvalds and the Linux team with a needs analysis to add the functionality required to exceed current enterprise systems. By telling the world what Linux didn't have to make it stand up to conventional UNIX, the development team went to work immediately and have many enterprise functions already in Beta testing.

In my opinion, Linux will move in two directions to capture the enterprise. The Linux development team styled their O/S after UNIX by using a similar organizational structure for files and directories, functions and commands. Analysts consider those similarities enough to call it a "clone". Analysts do little justice to Linux by considering it a clone. The UNIX style of Linux makes it familiar enough for the very senior technical people in companies to deploy it without having to take training on a new OS.

The significance of D. H. Brown's findings allow us to understand how the Linux community can exceed the thresh holds of performance of enterprise systems. While IBM AIX rates as the best among enterprise system, it only ranks as very good, not excellent. None of the conventional O/S platforms ranks excellent.

That's how an enterprise enhanced Linux begins to crowd other O/S platforms out of the way. Linux developers tend to produce a higher quality of application than conventional ones. Many reasons exist for this fact. Trying to understand why doesn't make sense to people who don't absolutely love technology. Suffice it to say their desire to exceed the present functionality of similar products and make themselves more valuable in the market place motivates many.

As the D.H. Brown reports breaks, people within the Linux community see some serious irony. Commercial packagers of Linux and ISVs have looked at Linux more in terms of a desktop to compete with Microsoft Windows and Apple Macintosh. Numerous development projects exist for this purpose. Projects such as KDE and GNOME complete favorably with other desktop interfaces in the market. In addition, computer users express excitement about Linux. Distribution channels exist for the product and it works on most computers even ones considered heretofore obsolete.

So what can we conclude? First, since Linux fits nicely between the desktop and the large-scale operating systems, it's really gotten its foot in the door. I suggest it will penetrate into both ends of the OS product market as its development continues. People want a seamless operating system that fits the end user that only wants e-mail, a wordprocessor and some sort of personal information management system. The people who run the enterprise want performance and freedom from all those rights management issues that come with licensing. Linux provides them both.

For my computer systems to work together, I need to use computers using the same protocols. Who believes that using a client on the desktop with a different protocol than the servers that run my network makes sense? Unwittingly, the solution appears within reach.

*D.H. Brown Associates, Inc. (DHBA) is a leading research and consulting firm that provides strategic analysis, assessment, and evaluation of technologies, products, and market trends in the Information Industry. DHBA concentrates on thoroughly understanding and continuously updating its knowledge of the core technologies and trends underlying current and future products. http://www.dhbrown.com

A copy of their report can be downloaded in pfd format from their web site.