The last couple of years have been a wild time for Linux. But, when you peel away all the hype, what progress has Linux really made in the marketplace? And, where is it going from here?
What does the crystal ball show for Linux? Now that the breakneck hype of 1999 has subsided, the free OS has some serious work ahead of it. Linux right now is moving into a dizzying array of new places — from watches to scientific cluster farms, to the everyday computer desktop, and while the industry seems to agree that it holds great potential, there remains a lot of work to be done before the penguin really does change the world.
In the beginning of 1999, a team of number-crunchers from the tech industry analyst firm IDC wrote up their annual four-year forecast for the operating systems market. They predicted that 1999 would be a good year for Linux and wrote that if the OS sustained the kind of growth they expected, it would surpass NetWare, gaining the number two position behind Windows 2000 by the year 2001. Not a bad showing really. But, a year later when the company again compiled its report on server OS marketshare, they found that Linux had already surpassed NetWare in 1999. “It happened a little more rapidly than we thought,” remembers IDC Program VP of System Software Dan Kusnetzky.
IDC certainly wasn’t the first company to underestimate Linux, and Kusnetzky says that IDC’s methodology, which only counts software shipments (not free copies or downloads), probably continues to underestimate how much Linux usage there actually is. But, by anyone’s reckoning, Linux has an incredible amount of momentum as a server operating system.
Today, a growing group of independent software vendors (ISVs) and system vendors are supporting Linux as a standard part of the Internet infrastructure. As the Internet becomes the medium for more and more services, a standard, inexpensive workhorse OS (like Linux), which can be deployed in great quantity at low cost, becomes more and more appealing. This has translated into exponential sales growth for well-known Linux systems vendors like VA Linux. It is also causing vendors like IBM, Compaq, and Dell to increase their support of Linux, with the hope of competing for the growing dot-com and Internet infrastructure market.
But, as more and more stakeholders come to the Linux party, questions are being raised about Linux’s community-driven development methodology and its ability to deliver the enterprise-level features that more and more vendors would like to see in Linux — things like 64-way SMP (Symmetric Multiprocessing) and NUMA (Non-Uniform Memory Access) — without re-creating the Unix wars of the early 1990s.
“I think that the biggest challenge is maintaining respect for our elders,” says Red Hat CTO Michael Tieman. “We’ve got a younger talent base than any industry I know…except gymnastics.” And while the relative youth of the Linux community is one of its great strengths, it can also create a myopic view of what is and is not important to the community at large. “It is very easy to presume that just because you can install Linux, you are a Linux expert,” he observes.
This is the whole challenge for Linux in the enterprise. How does Linux grow up without compromising the very principles that made it great? Fed by a youthful community of developers with strong university roots, Linux is currently nurturing a number of companies, many of whom see their best chance for success in the enterprise.
With virtually every significant Linux developer now employed by a company with a commercial stake in the success of Linux, will Linux be able to meet the demands of new users and partners while at the same time retaining its traditional openness and spirit of cooperation?
This question cuts both ways –how will commercialization affect Linux’s development and, on a broader scale, how will open source ultimately affect the enterprise? Most pundits agree that, for now, the enterprise is not prepared for open source. As with the personal computer revolution, corporate IT types are adopting Linux only when forced to by their sales or marketing departments, which may have Internet services that were set up by a Linux-savvy administrator. Before Linux can take over the corporate world, a number of things need to happen.
THINGS THAT GO BUMP IN THE NIGHT-
What Worries Linux’s Leaders?
“The ISV problem worries me a lot, and the LSB. It has to be the case that when a Linux vendor ports to Linux, they port once.” — Larry Augustin, CEO, VA Linux Systems
“The thing I’m most worried about is the encroachment of the intellectual commons, which is the effort to patent everything and to copyright everything. Linux is where it is because we have access to all the brilliant work that came before us.”
– Bob Young, Chairman, Red Hat
“Right now, I am very concerned about some of the larger companies who are coming into the Linux space. Their upper management kind of understands Linux — maybe they have one or two managers who really understand it — but it hasn’t quite trickled down through the entire organization.”
– Jon “Maddog” Hall, Executive Director, Linux International
“What worries me is software liability and warranty requirement legislation like UCITA [the Uniform Computer Information Transactions Act]. I think it’s fine for producers of commercial software to have some legal liability for the performance of that software, but people who give away their software can’t afford to be liable.”
– Bruce Perens, President, Linux Capital Group
According to IBM VP of Software Strategy Robert LeBlanc, CIOs are first trying to understand the value proposition of Linux — what it can do and what it cannot do, and most importantly, how Linux can be integrated with the rest of their IT operations. But beyond the rather mundane challenge of simply understanding Linux’s capabilities, there is a certain degree of discomfort that CIOs must overcome with the whole open source phenomenon.
“Open source actually scares them,” says LeBlanc. “Of course, the reality is there’s more structure and process in the open source community than there is in a lot of IT shops I’ve visited.” CIOs are risk-averse, says LeBlanc, and to many of them open source code simply represents a new risk.
But Linux is unlikely to completely mimic commercial development structure any time soon. If fretful CIOs are to have their concerns assuaged, it will not come from changes to Linux so much as from the widespread acceptance of open source as a valid and safe software development methodology. Today, developers can easily cite best-of-breed server software like Apache and Sendmail, which have been developed using open source rather than proprietary methods. And, with the help of open source service companies like Collab. net and VA Linux’s SourceForge Services, established commercial vendors are now beginning to open source some of their own software.
So, how will Linux penetrate the enterprise? “Tentatively and slowly,” according to IDC’s Kusnetzky. “In studies we’ve done as recently as the first of this year, over 80 percent of decision makers said they were open to Linux, but only 23 percent of them said that they had a Linux system in shop.” Why? Kusnetzky says that it’s because Linux still lacks the infrastructure necessary to make CIOs comfortable choosing it. He says that Linux needs applications, development environments, middleware tools, and a pool of certificate-in-hand Linux-certified administrators before it will make serious inroads into corporate IT –things that Microsoft and traditional Unix companies have spent years developing.
Building Blocks For the Linux Desktop
Loki Entertainment Inc. is best known for proprietary software. It ships high-profile games like Civilization: Call To Power and Myth II: Soulblighter. However, the company has also developed a lot of LGPL code — including OpenAL (a 3D audio API) and SDL (a multimedia API) — to improve Linux as a desktop platform.
Loki developer Sam Lantinga began working on the Simple DirectMedia Layer (SDL) multimedia library in 1996 while working on a Win32 port of the Macintosh Emulator, Executor. His vision was to provide a cross-platform API for things like accessing the screen, mapping mouse input, and playing sound.
Four years later, SDL has raised the bar for Linux desktop developers, providing sophisticated DirectX-type functions like streaming stereo audio, 3D graphics support, and hardware-accelerated video on a variety of OS platforms (including Linux, Win32, Solaris, IRIX, and FreeBSD).
But, Loki isn’t the sole developer of SDL. The multimedia library is itself an open source project and is constantly being developed by a “very active community,” according to Lantinga.
For more info, check out http://www.libsdl.org.
Move Over Mac
When it comes to desktop operating systems, it’s still very much a Windows world. And though IDC expects Linux to surpass MacOS as the second most popular desktop OS by 2001-2002, it does not expect Linux to make significant inroads against Windows any time in the foreseeable future. “The applications get in the way,” says Dan Kusnetzky. “If you look at the Windows environment, they have thousands of applications.” And, most importantly, Windows has such crucial desktop applications as Microsoft Word (95 percent market share), Excel (94 percent), and Outlook (80 percent) — none of which seem likely to be ported to the Linux platform before the Department of Justice forces some radical changes on Redmond.
But that is not to say that Linux has conceded the desktop market to Microsoft. Far from it. Some of the most exciting Linux development work happening today is aimed squarely at putting together the pieces of a viable, user-friendly, desktop Linux platform.
|EAZY ON THE EYES: With the release of their Nautilus file manager, Eazel hopes to create a new kind of user interface for Linux.|
One of the original developers of the Macintosh, Andy Hertzfeld, has made the creation of a next-generation computer interface for Linux his mission in life. He has co-founded a company, Eazel Inc., whose recently-released Nautilus file manager is just the first step toward creating a whole new kind of service-ready, Web-aware user interface — one that Hertzfeld and his team hope will spur the next generation of killer desktop applications.
A year ago, the Eazel team considered merging with another innovative desktop-centric company, called Helix Code, which is currently developing a consumer-friendly, distributed desktop environment which is based on the GNOME (GNU Object Model Environment) windowing system.
Another company that is betting on a viable Linux desktop market is Tustin, CA’s Loki Entertainment Software, which has spent the last two years porting a number of the top Windows games to Linux. The company’s CEO, Scott Draker, agrees that it is lack of applications more than anything else that is holding Linux back on the desktop. For technical users like software developers, he says, Linux is already “useable and quite feature-rich.” He adds, “If I were going to buy a computer for my mom, however, would I buy a Linux machine? No.”
But, applications are not the only weak link for Linux. Draker’s Linux wish list includes: video hardware, more and better drivers, easier installations, the general availability of Linux systems in retail outlets like CompUSA and Circuit City, as well as some kind of meaningful Linux standard. This last point — the need for all of the various Linux vendors to finally agree on things like what configuration files should go in which directories, and just which libraries the distributions should be using — is a popular rallying cry for almost every Linux company.
And yet, just as most Linux vendors stress the importance of the Linux Standards Base (the organization that is attempting to create just such a standard), they equally decry its lack of progress. Though the LSB has been working on defining a standard for over two years, it has yet to produce one.
Loki’s Draker believes that as larger companies begin to pour more resources into Linux, and as the size of the user base grows, the combination of customer demand and corporate desire will eventually force the issue and result in some kind of meaningful definition of the Linux standards.
“People want to have the LSB right now,” says Linux creator Linus Torvalds, “but people are not necessarily committed enough to put resources into it. So it’s not going very fast.” For Torvalds, however, the lack of progress is not a major cause for concern. Standards like the LSB should not be seen as “the road to heaven,” he says. “If you look at standards that way, you’re always going to be disappointed.” Torvalds believes that the community will eventually coalesce around the best technology, and that the role of the LSB is to follow behind, documenting what has already become de facto industry standard.
Desktop Market Share
Linux is expected to pass the Macintosh as the second most popular desktop OS by 2003.
| 1996 || 1999 || 2003|
|Windows || 78.5% || 87.3% || 88.9%|
|MacOS || 6.8% || 5.0% || 5.2%|
|Linux || <1.0% || 3.9% || 5.5%|
|Source: IDC, 2000|
Server Market Share
In terms of units shipped, IDC predicts that by 2003 it will be a horse race between Windows 2000 and Linux.
| 1996 || 1999 || 2003|
|Windows || 23.8% || 36.8% || 44.5%|
|Linux || 6.0% || 23.8% || 36.7%|
|NetWare || 29.9% || 18.0% || 9.5%|
|Unix || 18.6% || 14.8% || 8.9%|
|Source: IDC, 2000|
In the Palm Of Your Palm
The embedded OS space has proved to be one of the most exciting and anticipated growth areas for Linux. Virtually unknown as an embedded platform two years ago, Linux is now the focus of a number of embedded tools and professional services companies. And, though its share of the total embedded device systems market is still negligible, the free OS is beginning to appear in a number of machines, telephones, on-board automobile computers, and dedicated Internet appliances, such as firewalls and authentication servers.
The appeal of Linux is not simply one of price. According to LinuxDevices.com Executive Editor Rick Lehrbaum, most device makers are not simply looking for a free OS to put on their devices. They are willing to work with a Linux company like Red Hat or Lineo, and even pay royalties, mainly because Linux gives them choice.
“Developers and companies don’t want to be held hostage to their OS supplier,” says Lehrbaum. “With Linux, you have a dozen outside services or suppliers you can work with; or, you can roll your own.” And, unlike the desktop space, there is no one dominant supplier in the embedded space. According to Lehrbaum, the largest supplier of embedded operating systems, Wind River Systems, has only captured between 20 and 30 percent of the market.
Embedded Linux is still in the very early stages of development, however. Though the number of vendors of embedded Linux systems and services has mushroomed in the last year, the actual number of shipping devices using embedded Linux remains “on the thin side,” according to IDC/FTI senior analyst Paul Zorfass.
“Linux just doesn’t have the years of development that something like VxWorks or even a Windows NT does,” Zorfass says. He adds that the acceptance of Linux may be further hampered by the relatively large footprint of the X Window System, and even by the GPL (GNU General Public License) itself, which some device makers view as a restrictive license.
Lineo, Inc. CEO Bryan Sparks says that the hesitation toward the GPL is largely a matter of public perception. “I’ve had customers ask me, ‘I’ve got an application that I want to run on top of Linux, do I have to publish the source?’ [The GPL does not require this.] There’s still a tremendous amount of ignorance about how the GPL works.”
Sparks adds, “Usually, companies that want to keep some parts of their system software proprietary are still able to do so under the GPL. It’s just that they often do not understand the legally correct way to mix the proprietary and the open. We end up educating device manufacturers.” He says, “We haven’t yet run into a circumstance where we haven’t been able to work around the GPL.”
Sparks agrees with IDC’s Zorfass that the X Window System is probably too big to make Linux a major player in the PalmOS and Windows CE-dominated world of Palm-like devices for the time being. But, this has not prevented Samsung subsidiary Gmate from developing a Linux-based PDA called the Yopy. Based on a 206 MHz ARM processor, the Yopy will include Palm-sized versions of a number of Linux apps (a word processor, e-mail client, and personal information manager). Gmate hopes to piggyback on Linux’s advances as a multimedia system by including features such as MP3 and video players in the PDA, thus providing an X-based PDA that competes with PalmOS and Windows CE devices.
Will the Yopy succeed? Will Linux take the desktop? Will it become the standard Unix that the IBMs, and Suns and HPs, of the world were unable to create ten years ago? At the end of the day, skepticism about Linux really amounts to skepticism about the open source model — it’s scalability and its ability to combine the efforts of people with different interests in creating a common, useful infrastructure.
Open source doubters believe that Linux will eventually collapse under the weight of its own development methodology. But as far as the open source faithful is concerned, we ain’t seen nothing yet.
Think Linux can’t run your wristwatch? Think again. The penguin has been spotted powering the following devices:
WRISTWATCH: Researchers at IBM’s Thomas J. Watson Research Center have created a prototypical Linux wristwatch called the WatchPad.
GOLF CART: ParView, an on-board computer for golf courses developed by Applied Data Systems and ParView, Inc., is now being used to give golfers GPS-aware course information and scoring.
TELEPHONE: As part of a partnership with Red Hat that promises to bring Linux to several Ericsson products, the mobile phone maker is about to release a Linux-based computer phone called the HS210 Screen Phone.
GAME CONSOLE: An Open Source Sega? Why not? Nine-month-old Indrema Corp. is developing a video game Linux distribution called DV Linux. The first DV Linux product, the Indrema L600, is expected in the spring of next year.
RADIO: It looks like a radio but connects to the Internet like a PC. Two companies, Penguin Radio and Kerbango, are developing Linux appliances designed to bring the world of Internet radio to consumers for the cost of $200-$300.
Though pretty much everything about the post 2.4 Linux kernel is in flux (including the name — it’ll be either 2.6 or 3.0, depending on what goes in it), there are a few key things that developers are targeting for the next major release of Linux. The highlights are:
- NUMA (Non Uniform Memory Architecture) Support
- A Journaling Filesystem
- 16-Way SMP
- Improved IDE Support
Robert McMillan is executive editor for Linux Magazine. He can be reached at firstname.lastname@example.org.