In late 2004 and early 2005, high-level matters dominated conversations and predictions about the open source market: the legal ramifications of using open source, how the community could establish an intellectual commons for germinating new technologies, and so on. In recent weeks, however, the discussion has taken a far more granular and competitive direction.

Long-established vendors such as IBM and Novell have announced or strengthened their open source offerings from the data server to the desktop, sometimes offering and supporting certified versions of open source applications, sometimes using open source code in new server-based configurations that let customers choose specific functionalities per user. New companies such as SpikeSource (http://www.spikesource.com/), a San Francisco-based startup offering integrated, validated, and certified open source stacks with ongoing maintenance and support services, are convincing venture capitalists the open source stack is a viable investment. Open source development houses themselves are being recognized either for delivering standalone best-of-breed technology—such as the Java application server JBoss—or as valuable acquisitions in a larger vendor's portfolio. In early May, for example, IBM purchased Gluecode (http://www.gluecode.com/), a well-regarded vendor of an integrated Apache Geronimo-based development stack that analysts believe will serve as an entry product for small and medium businesses or departments of larger enterprises that don't yet need full WebSphere capabilities or complexity.

Nick Halsey, vice president of sales and marketing for SpikeSource, says, "CIOs are saying, 'Pandora's out of the box. How are we going to support this, manage the deployments, certify these components work together, and know that when we implement a patch, it won't break the stack?' All the problems you've faced in any enterprise support environment exist in the open source environment, and yet there was nobody there fulfilling that need."

That hole in the market will not stay empty for long. Just as startups saw an opportunity to assist IT managers in making sure their stacks work, so too are the established powerhouses such as Novell and IBM. As the various players jostle for position, the open source community's perceived collegial camaraderie will no doubt suffer from some competitive body blows. As open source becomes a more viable business model, the stakes involved are higher. And while many early deployments of open source software, particularly Linux and the Apache Web server, were done far from the eyes of end users, the maturity of open source end user applications has lately provided a heightened element of visibility and passion to the debate.

"In terms of where the emotion in the marketplace exists, there is a lot of emotion and passion from various segments of the community around this issue called sometimes the Linux desktop," says Don Harbinson, program manager for Linux strategy at IBM's Software Group. "When we rolled out Lotus Domino Server on Linux, the companion question I was constantly being asked was, 'Where is Lotus Notes on Linux?' And our answer at that time, which was accurate, was that our strategy at that stage was to focus on the server side of Linux. We didn't position our answer as 'Never ever,' but the server side was where our focus was in 1999."

As open source end user applications have matured, however, IBM has joined other vendors in bringing them to the desktop in one form or another. These applications' visibility has spotlighted the sometimes prickly relationships among vendors that may prove problematic in future development.

In late April 2005, members of the OpenOffice.org (http://www.openoffice.org/) community publicly criticized IBM for basing the productivity editors in its server-managed Workplace Client Technology on OpenOffice code, but not contributing to the OOo effort. IBM, however, is trying to differentiate its offerings from the perception that Workplace's technology is an alternative office suite.

"Our intent was to serve customers' needs with our Workplace strategy," Harbinson says, "and we heard this very loud cry of pain around the cost to maintain and continue to update their users' desktop environment. That's where the costs are and that's well known and recognized in the industry. Workplace is aimed right at that, but we don't want to be an alternative to Microsoft Office or OpenOffice, because we exited that business."

The introductory documentation for the Workplace Client technology (pdf, http://www.redbooks.ibm.com/redpapers/pdfs/redp3947.pdf) further explains IBM's philosophy in deciding to offer server-based components of productivity applications instead of a full PC-installed suite. The authors restate the 80-20 rule, which holds that 80 percent of any application's users need only 20 percent of its capabilities, while just 20 percent need 80 percent.

Certain sets of users don't require a full office suite such as Microsoft Office. Their work may require them to view, and in some cases create or edit, basic Microsoft Office documents. This class of users is well served by the productivity tools found in the IBM Workplace Client Technology Rich Client Edition. The productivity tools support Microsoft Office document formats and enable users to read these documents. The productivity tools also provide editing capabilities that almost rival the feature set in a full office suite.

In some ways, the IBM-OpenOffice discussion summarizes the future points of open source contention in a nutshell. As end user technology changes and customers demand new delivery models, open source developers have to weigh the health of large and popular projects such as OpenOffice with the needs and desires of customers and vendors' abilities to supply variations on a theme. Yet too much customization at the expense of a parent project could kill the goose that lays the golden egg, resulting in updated forks and an obsolete parent.

James Governor, a partner at analyst firm RedMonk who follows IBM and open source work extensively, sees the good and the bad of the complexities of forking open source code.

In the IBM-OpenOffice case, he says, "Where we think it was almost necessary is in thinking beyond the monolithic desktop and componentizing it so you could have the component in a workflow. That's something that's potentially very valuable. It fits in IBM's portal strategy and makes a lot of sense. The flip side is, but it would have been nice if they hadn't forked it. Nice is a terrible word to use, but imagine the power of a proposition that IBM and Sun were working together on this rather than forking a code base. This question of localization will have to be addressed; you don't want to write the same code twice."

"Most people are realizing we are moving from 20th century technology—the keyboard, monitor, hard drive, the fancy typewriter, essentially—to something really predicated on the Internet," says Louis Suarez-Potts, community manager for OpenOffice.org. "Open standards and protocols are key to successful development for these new applications, so it's vital to be able to be ported to what users need and want."

And, as OpenOffice's technology matures, more and more users are becoming familiar with it. Suarez-Potts says the organization can track 40 million downloads since May 2002, with the majority in the last 18 months and a steady growth curve to 400,000 to 500,000 a week. After the highly anticipated release of OpenOffice 2.0 final version, expected by June depending on debugging requirements, Suarez-Potts says that number could reach 800,000 to 1 million a week. As users' needs migrate to more mobile devices, componentized open source editors such as those found in Workplace might be vital offerings from OpenOffice; yet because Workplace is derived from OpenOffice but still not part of OpenOffice, an open source OOo component would have to be developed from scratch.

Governor's partner at RedMonk, Stephen O'Grady, maintains a Web log (http://www.redmonk.com/sogrady/archives/000643.html). In it, he says that open source fans might well be disappointed that IBM isn't an OOo contributor, but he isn't about to issue a clarion call for the company to reverse course, for two reasons: IBM has the legal right to pursue the path it is taking, and in other areas the company is a willing and, in many cases, dominant contributor to open source projects.

One such open source application IBM has embraced in its totality is the Firefox browser (http://www.mozilla.org/products/firefox/). The company contributes to the Mozilla Foundation, and in early May announced it was going to offer internal Firefox support for its employees.

Governor says the strategy is logical for IBM, one that it pioneered with using open source in servers and middleware and is now expanding to the end user.

"They want to make an environment users feel very comfortable in, and once they become comfortable they can slide Windows out and slide Internet Explorer out. For the end user, you'll be in the position where you're getting the same experience but not paying the margin to Microsoft."

Governor's observation regarding the huge cost savings possible with open source software recalls an issue that has long tantalized and tortured the enterprise IT community. On one hand, open source offered no licensing fees; on the other, it also offered little in the way of support. This lack of support was a key selling point for proprietary vendors, often cited as TCO, or total cost of ownership: a free application that didn't run, or took too many resources to run, cost more in the long run than a proprietary application that was supported. But the open source support infrastructure is finally catching up with the promise.

"You have several companies that are building their own LAMP [the open source Linux-Apache-MySQL-Perl/PHP/Python] distributions, and they're going to be in the business of providing reliable support for them," says independent analyst Amy Wohl. "That seems to be interesting enough that each of the companies that's decided to do it has managed to get significant amounts of venture capital, which I find quite interesting since a few years ago 'open source' and 'venture capital' were never said in the same sentence."

In fact, SpikeSource was hatched inside one of the most prominent venture firms of the Internet boom, Kleiner Perkins Caufield & Byers (http://www.kpcb.com/). Murugan Pal, founder and chief technology officer at SpikeSource, was an entrepreneur-in-residence at the firm in May 2003, when he and former Oracle president Ray Lane began laying the foundation for SpikeSource. They talked to 30 of the Fortune 50 CIOs, among other experts, and discovered that while open source was appealing, the executives also had three main areas of concern: interoperability; support (70 percent of the world's Web servers are Apache but many run without support, Pal says); and policy, governance, and usage. Nearly half the people interviewed two years ago said they would sign up for a service like SpikeSource's on the spot if it were available.

SpikeSource integrates approximately 70 open source components into its Core Stack, which allows developers to customize according to their needs. The company also offers Spike Asset Manager, which audits architectures to ascertain which open source applications are running. Pal and Halsey say their automated services allow customers to focus on their business instead of worrying about how well their open source software is running. Their services also give customers more visibility into the best emerging open source applications. SpikeSource's Halsey says the company provides a vendor-neutral perspective that can assure customers they're using the highest-quality software, with no constraints locking them in to a single vendor's thrall.

However, the traditional vendors have noticed the likely viability of the open source assurance business and are not leaving it to the start-ups.

"I think companies like SpikeSource are filling temporary voids," says Ed Anderson, vice president for product development for Novell's Platform Group, says. "As traditional software companies, big companies who own the customer relationships today, bring open source offerings into their portfolio, over the long term it will obviate the need for these other service companies."

Novell, which has thrown its support strongly behind open source architectures, announced two such support programs at its BrainShare conference in March. The Novell Validated Configuration Program focuses on defining and certifying integrated, multiapplication and platform stacks on Linux. The Novell Market Start Program will help open source vendors develop a marketing plan that defines the best way to reach customers and to leverage Novell's sales channels.

Analyst Wohl says anybody with a stake in open source deployments should watch developments carefully as the offerings shake out. She predicts more acquisition action, which could mean a realignment of strengths and liabilities in the vendor community. For example, she says, IBM's Gluecode purchase may put JBoss into play.

"How long might it be before Hewlett-Packard or Computer Associates or Sun buys JBoss?" she asks. "Because these things never happen as single events. They always happen as the first act of a multiact play. I have to assume we're going to see some more activity here."