Once a purveyor of closed and proprietary solutions, the company has recast itself as flag bearer of all things open and interoperable. Seeing the company's peace, love and penguin ads for Linux are indeed worthy of an eye-rubbing double take.
IBM officials speak fondly of an attitude adjustment that has changed Big Blue into a much more customer-focused company. They describe an altruistic metamorphosis that benefits end users who will settle for nothing less than interoperability and support of open standards.
But is there more to the picture than meets the eye? Are IBM's interoperability efforts and open source donations such as Eclipse and code for UDDI, SOAP, and Linux subtly serving something more strategic for Big Blue?
Linux does three things for IBM. First, Linux presents IBM with an opportunity to offer buyers a reliable, scalable, and relatively secure Intel-based alternative to the server versions of Microsoft's Windows. While IBM itself doesn't offer a Linux distribution (SuSE and Red Hat do that), it no doubt sees Linux' potential to reduce the company's dependency on Microsoft for sales of Intel-based servers. Second, Linux turns a high-end Intel-based server such as what IBM introduced at CeBit into a low-cost alternative to Sun's low-end servers. Third, Linux creates a migration path for IBM's customers to easily move off the Intel platform and onto any of IBM's bigger iron. Linux is now available on just about all of IBM's systems, right up to the mainframes. If customers running Linux-based, back-office applications like Oracle 9i or IBM Lotus Domino server want the reliability and performance advantages of a mainframe without the pain of an operating system lobotomy, IBM can give it to them.
There's no question that IBM's involvement in Linux is, in some ways, a power play against Microsoft, Intel, and even Sun. IBM could take much more control over its destiny if it minimized its dependency on these companies. But in order to strategically marginalize these three companies, IBM would have to place a few more bets under the industry-friendly guise of open source.
One of these would be IBM's donation of code that makes it easy for Java developers to use the Simple Object Access Protocol (SOAP), a specification for formatting XML-based remote procedure calls (RPCs). At first glance, the donation made IBM look like a good open source citizen. But now, almost two years later, the donation is one piece in a power play that could end with the control of Java being wrested from Sun.
SOAP was invented by Microsoft and, shortly thereafter, got a facelift with the help of IBM. In April 2000, a new and improved version 1.1 was submitted to the World Wide Web Consortium (W3C). SOAP's main claim to fame was then--and remains--its platform independence. It enables any system to invoke an RPC on any other system, regardless of the operating system or language. This sort of lingua franca does more than make it possible for heterogeneous systems to send or reply to RPCs. It makes it easier for heterogeneous systems to replace each other.
According to IBM's Director for eBusiness Standards Strategy Bob Sutor and Microsoft's .Net Platforms Strategy group director Neil Charney, Sun wasn't thrilled. The fact that IBM, at roughly the same time, tried to launch an organization (OpenServer.org) whose goal was widely recognized as "supersetting Java" (see "When will IBM buy Sun?") couldn't have helped matters. If OpenServer.org put a crack into Sun and IBM's relationship, and SOAP drove a wedge into that crack, then what came next probably blew things wide open. Two months after the SOAP spec was submitted to the W3C (well before any standard was ratified), IBM open-sourced all of the Java code necessary to make use of it. Regardless of what the W3C did, a de facto standard had been signed, sealed, and delivered.
At the same time, another standard that IBM and Sun had collaborated on--ebXML--was about to become the center of controversy. Throughout 2000, while the two companies worked together famously on ebXML for the United Nations and explored the idea of XML-based registries, IBM was also working clandestinely with Microsoft and Ariba on another directory specification, Universal Description Discovery and Integration (UDDI).
ebXML proponents and companies like CommerceOne, who had committed significant resources to the registry work being done with ebXML, were stunned when they first learned of UDDI. But the sequence of events demonstrates how much pressure a move like this by Microsoft and IBM can create on the industry. The forum at ebXML.org (a part of OASIS) shows CommerceOne's Bob Glushko expressing surprise and frustration on the day the UDDI story was released in the New York Times. Then, one day later, CommerceOne is listed as being a supporter of the project.
A pattern was beginning to emerge. To many, including Sun, it must have felt like the second blow of a one-two combination. The UDDI Project was born on Sept, 6, 2000, and Sun, having just seen SOAP steam-rolled into a standard, probably felt like it had a gun to its head and signed up as a supporter. Anybody might have guessed what would happen next: A few months later, IBM contributed its Java-based UDDI code to the open source community. If Sun, the Java Community Process, or the other ebXML supporters had any objections, it wouldn't matter. When code like this is given away, it has the same effect on developers that the price of Microsoft's Internet Explorer (free) has on end users. A standard is born.
Open-source, widely viewed as a way for the development community to participate in the evolution of software that's owned by nobody but shared by everyone, was now a competitive weapon and a means to an end of IBM. (See "When will IBM buy Sun?") But IBM wasn't done. In a move that would enrage Sun, IBM open-sourced all of its Eclipse code with a contribution valued at $40 million. If SOAP opened the superhighway for RPCs, Eclipse would be the developer's interoperability interchange between target operating systems. Eclipse technology makes it possible for developers to target Linux, Java, or Windows with one integrated development environment. By virtue of their Eclipse support, the tools that plug into the IDE offer a degree of platform independence that their Windows, Java, or Linux-specific sisters can't. Most viewed the move as a way for IBM to get Windows developers to start targeting Linux. Sun saw it as a threat to the tenants of the Java Community Process and the write once, run anywhere promise of Java.
Once again, IBM had played open-source perfectly to its advantage, leaving only one question. What's next?