ARMONK, N.Y.-IBM Senior Vice President and Software Group General Manager Steve Mills met with three of his top lieutenants in his office back in 1997 to discuss the “Webification” of IBM’s enterprise tools. Out of that discussion the IBM WebSphere Application Server was born.
In an interview here at IBM headquarters, Mills told eWEEK how he and Danny Sabbah, John Swainson and Alfred Spector met in Mills’ office in Somers, N.Y., to discuss IBM’s future in the nascent application server space when the market competition was growing hot and heavy. Only Sabbah remains at IBM of the three lieutenants. Sabbah, who was Mills’ CTO at the time, is now general manager of IBM’s Rational business unit. Swainson is CEO of CA, and Spector is vice president of research and special initiatives at Google.
Mills recalled that at the initial meeting a discussion emerged about which HTTP server to use. “Besides a Web browser you also needed an HTTP server stack,” Mills said. And while IBM had one of its own, “the most popular was the Apache Web server, and we made a decision to anchor our effort to Apache because it had 47 percent market share,” he said.
Back then the competition in the application server space included companies like Bluestone Software, Silverstream, NetDynamics, Kiva Software, Allaire and WebLogic, which was an independent company then.
Following that initial meeting in his office, Mills sanctioned a small team of about 25 people in Raleigh, N.C., to begin working on the technology that was to become WebSphere.
Chris Wicher, who is now vice president of Mid-Market Solutions in IBM’s Software Group, was the original WebSphere product development executive back in January of 1998 when the team began working on the technology.
“When we started WebSphere, the first two releases-in the second quarter of 1998 and the third quarter of 1998-were focused on rapid development and deployment of Web applications supporting HTTP, Servlet and JSP [Java Server Pages] apps,” Wicher said. However, “IBM quickly extended WAS [WebSphere Application Server] to transactional applications and beyond, driven by customer needs, and anticipating where companies would want to have extensions to gain more value to their business.”
Describing the initial setting, Wicher said, “We started WebSphere in the first quarter of 1998 at the request of Steve Mills, with a small team of IBMers who knew the emerging Web technologies and what customers were trying to do. … Within six weeks of starting the project, we went back to Steve Mills with a prototype. He quickly absorbed what we had and where we were headed, gave us guidance on leveraging the Apache Web server, created the name, and asked us, ‘How fast can you ship this as a formal IBM product?'”
Wicher said his team worked at a furious pace to deliver the first generally available WebSphere release.
“This cycle time for a formal product-from prototype concept to GA in less than four months-was fairly revolutionary in IBM in those days, so everyone on the team assumed multiple roles,” Wicher said.
“The managers were working managers, including writing docs, testing, working with customers and doing the product planning. The ‘developers’ did it all … design, coding, testing, quality certification, customer betas, pubs, etc. It was like being in an atmosphere where the team was on a quest-high energy, everyone passionate about delivering a real product to market which customers really liked (on an unheard-of schedule).
“The team turned right around and put out a second release three months later. Again, the energy, the ‘quest passion’ made all obstacles seem insignificant. This is not to say that people were not tired. In this time frame, the managers were bringing in pizzas every night to the team. And the days and nights were long, but the team-generated energy was contagious,” he said.
However, despite delivering two versions of WebSphere in 1998, “by the end of the year, customers were saying, ‘Our problem is not totally solved,'” Mills said. IBM customers said they wanted their transaction monitoring software and their component broker technology to work alongside their application servers, he said. “They said these three things are different and life could be easier if you gave us one thing.” IBM had acquired Transarc, which was the maker of the Encina transaction monitor-which also is where Alfred Spector came from as founder and CEO of Transarc.
So IBM set out to deliver on this request and by 2002 had delivered all the functionality customers were asking for while also extending the technology, Mills said.
Ambitions for WebSphere Expand
In a case study he wrote for his students, Professor Chris Trimble of the Tuck School of Business at Dartmouth College said, “The WebSphere application server had gained considerable momentum by 2002. IBM released new versions every year. Standards were evolving quickly, and that created complexity. IBM inevitably created some uniqueness in its software but linked to external standards wherever possible.”
Trimble, who spent time in Somers studying WebSphere, also said, “As of 2002, IBM’s strategic objective for WebSphere was straightforward: to offer the most capable application server on the market. The key criteria buyers would evaluate were the number of systems with which the server could connect (e.g., IBM databases, IBM mainframes, Oracle databases, SAP enterprise software, Siebel [CRM] customer relationship management software, and so forth) and their security, speed, scalability and reliability. Also, buyers evaluated how easy it was to develop new applications to run on the application server.”
But by 2004 the ambitions for WebSphere expanded and other IBM software brands contributed to the family of WebSphere products
“As WebSphere expanded from its core product, the WebSphere application server, to a broader tool set for programmers, the WebSphere brand also expanded, from a certain set of features to a philosophy of how modern corporate IT systems should be built and managed,” Trimble said. “One core tenet of the philosophy was that applications developed using WebSphere should be interoperable with most any system and easy to reuse.”
In an e-mail exchange with eWEEK, Trimble added that other factors leading to the success of WebSphere included, “The senior management team was directly involved, in a critical way, by closely managing the interactions between nascent businesses like WebSphere and the rest of the company, ensuring that the WebSphere could leverage IBM’s massive assets without getting destroyed by quarter-to-quarter hit-the-numbers imperatives. And IBM invested steadily in WebSphere over 10 years, even as the rest of the industry went through the dot-com boom and bust.”
Meanwhile, IBM continued to deal with stiff competition from products such as BEA’s WebLogic. However, “BEA never really embraced a true next-generation design. Their design point was simpler and more vulnerable to attacks from below,” Mills said, referring to offerings such as the open-source JBoss application server.
Mills said WebSphere is in the same class as core IBM technologies such as the mainframe-based IMS (Information Management System) database and the CICS (Customer Information Control System) transaction server.
“In Java application servers, I consistently see WebSphere, Oracle WebLogic (former BEA) and JBoss,” said John Rymer, an analyst with Forrester. “Sun GlassFish is a challenger to JBoss’ dominant mind share among open-source options that clients are starting to bring up.”
Craig Hayman, vice president of WebSphere in the Application and Integration Middleware Software Division of the IBM Software Group, said he has been working on WebSphere since its inception 10 years ago.
“In the early days we took WebSphere from an idea to a product, then from a product to a platform, and then from a platform to an SOA [service-oriented architecture] portfolio,” Hayman said.
Bulletproof or Threatened by Open Source?
Mills said evolution has made WebSphere bulletproof. He said he is “not particularly concerned with competition” in this space, particularly from open-source offerings.
Yet, Forrester’s Rymer said, “I think open source is a problem for IBM and Oracle WebLogic. As our quality survey suggests, the open-source alternatives are ‘good enough’ options for many shops for a lower cost than the conventional products. JBoss and Sun are benefiting from this market dynamic.”
Despite its having been built on top of the open-source Apache Web server, Mills said there are no plans to open-source WebSphere. “Something of this class of software could never be free,” he said. In the mainframe world, IBM has delivered software as source code, but that is not likely to occur with WebSphere, Mills said.
Moreover, WebSphere represents a good illustration of how IBM evolved to a much more collaborative development model, Mills said. The initial development team in Raleigh eventually branched out to include developers in Austin, Texas, and Pittsburgh and then a dozen locations, Mills said. Today WebSphere is developed in 80 locations by 6,000 developers, Hayman said.
What’s next for the technology?
“Scale, scale, scale and more scale,” Mills said. “We’re going for more data, more transactions and more performance. We’ve improved automation, self-diagnostics and enhanced recovery. We built WebSphere with a very weak operating system in mind, so WebSphere had to be very operating systemlike. We built WebSphere with the expectation that it would be running on Windows or Linux or something else, not MVS.”
Rymer wrote a report for Forrester saying that application server users can expect more SOA, social computing, RIA (rich Internet applications) and Web 2.0 technology in their application servers. And that is exactly what IBM is delivering.
The next version of IBM’s solution, WebSphere Version 7, will ship later in 2008. Hayman said IBM has been building around six basic points to improve WebSphere: service orientation, analytics, active content, business policies, business rules and events.
“Version 7 will enable you to do more work with fewer servers,” Hayman said. “You will be able to use fewer machines to achieve more work-the computing equivalent to improving miles per gallon.”
In addition, WebSphere 7 will include improved management capabilities and enhanced Web 2.0 support in the form of support for REST (Representational State Transfer) APIs and for the Dojo Toolkit, an AJAX development tool kit.
“What WebSphere has become today is a run-time [for running] all kinds of workloads-from J2EE [Java 2 Platform, Enterprise Edition], to REST, to Spring, to Web 2.0, etc.,” Hayman said.
Beyond Version 7 of the technology, Hayman said IBM will focus on delivering more support for events, governance and business process management.