The Importance of Linux

- by Paul Murphy -

The personal computer industry is about to arrive at a "tipping point" in which everything changes and those who get ahead of the curve earn social and cultural advantages over those who don't.

A social or economic tipping point is characterised by a sudden burst of mass sanity as mainstream public opinion abandons an unsustainable mythology in favor of something closer to reality. Such reversals usually have a sustaining mythology of their own, complete with an acceptable explanation for previous behaviour. In the case of the PC industry tipping point now upon us, that mythology is built around the legend of Linus Torvalds.

According to the legend Mr. Torvalds, a poor graduate student in Finland, single handedly invented a computer operating system called Linux along with a free public distribution method called open source and then used this Robin Hood combination of product and method to stand tall against the forces of evil in the form of Microsoft, Intel, IBM and other huge international corporations involved in personal computing.

In reality the facts around which this myth evolved were generally right, but the perceptions built around them almost completely wrong. Mr. Torvalds is Finnish, did start work on Linux as a graduate student, did freely distribute his work using the internet, and did motivate tens of thousands of very bright people to copy, use, and extend his work. As a result he really is an inspirational figure in the industry whose work has been pivotal in getting the world of small systems computing where it is today.

On the other hand Linux is no more than another expression of the core Unix ideas originated at what was then AT&T's Bell Laboratories by Dennis Ritchie, Ken Thompson and others in the early seventies. The open source movement is just a faster, internet enabled, implementation of the much older academic tradition of peer review and building on foundations laid by others. Furthermore, the Linux operating system itself is neither a new invention nor a stand alone product. It consists of a Linux kernel developed by Mr. Torvalds and his colleagues by radically improving an earlier open source Unix released by Andrew Tannenbaum in 1987, the Gnu utilities developed by the free software foundation, several graphical user interfaces akin to Microsoft's Windows brand products, and a slew of third party applications.

Torvalds himself has never claimed to be more than he is, but tipping points aren't made out of technical reality, they're made out of perceptional change. Thus it was the legend of Torvalds, not the reality of his actions as a kind of Wayne Gretzky of Unix development, that gave Linux the patina of political correctness needed for it to gain widespread public acceptance.

One of the things that gives Linux its power in the public mind is that comparisons have historically been to Microsoft's Windows brand products, not to other versions of Unix. On all reasonable measures of performance, stability, and technical accessibility, Linux is well ahead of the latest Microsoft products and so shines in such comparisons. It is not, however, remotely a leading edge system in the same class with the BSD family of Unix products and Sun's Solaris.

Apple, for example, uses a BSD variant called Darwin as the foundation for MacOS X: Unix my grandmother can, and does, use. Sun Microsystems, meanwhile, is evolving Solaris into a network based environment offering failure free computing to business users both in the data center and on the desktop. A Sun Ray user interupted at work can, for example, pull her java card from the machine she's working on, cross the country to another office, plug the card into a machine there, and continue typing where she left off.

Thus both the BSD and Solaris groups offer technical advantages Linux doesn't yet match, but that doesn't matter: technology advantages don't lead to social revolutions. Technology often empowers revolutionaries, but social change is driven by widespread acceptance, not functionality or value.

Linux acceptance has been driven by its perceived political correctness in the mass media, itself an artifact of the legend of Torvalds rather than technical value, but validated by cost and technology comparisons to Microsoft's products.

Notice that Linux runs on the same hardware for which Microsoft makes it products, thus corporate cost comparisons between Linux and the Microsoft PC depend totally on the software and have little, if anything, to do with the hardware.

Walmart, for example, sells a fully equipped Linux PC for $796 including a 15 inch LCD and lots of free software meeting the needs of office professionals, systems developers, and web service providers. In contrast, the comparable Dell's $799 2400 combines what amounts to the same hardware with only a basic Microsoft operating system license and no applications. Adding Microsoft Office costs $399 more, a code development system adds $529; even just extending the OS license to the office release adds $70 -and all of this stuff has free equivelents that are as as good or better under Linux.

To a business, however, operating costs are far more important than capital costs. Thus experimental business uses of Linux were legitimized by the legend and sold to management on the promise of capital cost savings, but second and subsequent installs tended to be for operating cost reasons. Fundamentally, everybody likes to be leading edge, particularly if doing so is politically correct and there are a few bucks to be saved in the process - but most of those who did replace a few Windows servers with Linux soon found that the software's quality led to much bigger benefits in terms of operational stability, support staffing, and the overall integrity of their information systems.

As a result Linux now dominates web services on Intel style computers like those Microsoft depends on and the contagion is spreading into areas such as file and print services or identity management where Microsoft previously had a lock on the customer. Equally importantly a number of companies, led by Sun Microsystems, have been developing open source desktop applications aimed at making Linux a competitor to Microsoft's Windows brand products for desktop service for home and business users.

Most people agree that products like Sun's Java desktop don't have as many features as Microsoft's integrated office suites, but people willing to give up some bells and whistles are finding the open source products fully functional and free of the proprietary limitations built into Microsoft's products. In effect, what they're finding is that an open source, package like the suite doesn't have all the features claimed by Microsoft's Office Professional, but in compensation it doesn't lose files, there's no cost beyond downloading and setup, no upgrade pressure, and few restrictions on exchanging files with other systems - including Microsoft's.

In consequence a growing number of organizations are experimenting with the Linux desktop and getting the same results they found with server functions: they save the expected dollars in up-front costs and surprise themselves with big gains in overall systems stability, support, and information integrity.

By themselves these trends are evolutionary, not revolutionary. If other things stayed the same, competitive forces would eventually raise the cost of Linux, lower that of the Microsoft alternative, and allow the market to reach a new balance between Unix and the Microsoft PC. Indeed companies like Red Hat have already been raising the cost of Linux support to just below that of a discounted Microsoft services contract but, fortunately or unfortunately, things don't stay the same.

In this case the triggering event that will push Linux over the tipping point and into desktop dominance is coming from way out in left field - IBM's work with Sony and Toshiba on a new CPU design known as the cell processor.

By all reports this new machine is an order of magnitude (10 times) faster for some key functions than the best Intel can offer. More to the point, the charter given IBM's newly formed PC design consortium for the Chinese and related markets, known as, strongly suggests that IBM has selected Linux over BSD as the operating system for this machine in both its desktop and server configurations.

The cell processor is fundamentally a GRID style super computer (consisting of many small computers connected in a grid pattern so they can work together) on one piece of silicon. Thus all the major Unix variants, including Linux, will move easily into the new computing paradigm because they all support GRID computing now. Microsoft's software products are, in contrast, almost wholly dependent on Intel's "x86 architecture" -meaning PCs built using chips, like the 8088 in the first IBM PC, itself derived as a downgrade from the original 1979 Intel 8086 processor.

Indeed Intel's own attempt to break away from the x86, its Itanium next generation CPU, has been a market failure in large part because it is incompatible with Microsoft's existing software and Microsoft has been unable to deliver significant new applications for it.

As a result the fact that Linux has traditionally been compared to Microsoft's Windows brand products and not the other Unix variants, will most likely lead the general public to perceive all this as Linux sailing on to new horizons while Microsoft stalls out - totally reversing the previous mainstream view that Microsoft and Intel were somehow at the forefront of high technology computing and thereby pushing Linux over the magic of a social tipping point.

Paul Murphy wrote and published The Unix Guide to Defenestration. Murphy is a 20-year veteran of the IT consulting industry.