NEWS AT SEI
This article was originally published in News at SEI on: March 1, 2002
Since 1996, the SEI has focused on software architectures in its Architecture Tradeoff Analysis (ATA) Initiative, based on two key premises:
- Desired quality attributes in a system such as performance, modifiability, interoperability, and reliability depend more on software architecture than on code-level practices.
- Quality attributes ultimately determine a system’s success.
SEI architecture practices complement the SEI’s well-known work in software process improvement by focusing on the design and quality of the product itself. A balanced product and process focus is important regardless of the development context—DoD contractor or “dot com” startup.
Jeromy Carriere and Steve Woods, former members of the SEI technical staff in the Architecture Tradeoff Analysis Initiative, are two of the founders of Quack.com, a startup company that was acquired by America Online (AOL) in August 2000. Their story and the story of the company they helped found demonstrate the broad applicability and utility of the SEI’s work in software architectures. Jeromy Carriere came to the SEI from Nortel in 1997. Early in his tenure at the SEI, Carriere was an integral part of the SEI team that supported development of the Control Channel Toolkit1 for the National Reconnaissance Office (NRO). He also led work at the SEI in architecture reconstruction and participated in defining the SEI’s Architecture Tradeoff Analysis Method (ATAM).
"It was this concentration on architecture and the attention to quality attributes that differentiated them from their competitors and elevated them to a place of prominence."
After completing a PhD in constraint-based reasoning for reengineering at the University of Waterloo, Steve Woods did post-doctoral work in Hawaii, where he first met Alex Quilici and began to explore ideas with Quilici for startup companies. Woods came to the SEI in 1998, where he immediately began to work on architecture reconstruction with Carriere and began discussing with Carriere some of the ideas that he and Quilici had been considering: initially, an Internet radio and other consumer wireless devices. As the discussions among the three continued, their ideas evolved first into a device that could aid a consumer during a shopping trip, and finally to a product that would enable speech recognition on a cell phone.
“Eventually it became clear that we should try to make a go of it,” says Carriere, “and that’s when we decided to leave the SEI. All through this, we were developing an architecture to support our ideas. In those days, I had been steeped in the SEI ideas about architecture, a highly structural perspective, but one driven by quality—looking first at what qualities my system has to achieve, and then making architectural decisions to achieve those qualities. Architecture is usually conceived from a functional perspective—what a system needs to do—and then you allocate functions to components, build those components, and then cross your fingers and hope that they work together; and then cross your fingers again and hope that they meet your reliability, performance, and flexibility goals."
“We were thinking of the qualities first. The quality that we needed primarily to achieve was flexibility. And that’s because we needed an architecture and a set of architectural principles that were going to be able to adapt quickly to changing market drivers and business goals; our architecture enabled us to be maximally flexible with respect to all of those things.”
Carriere and Woods attribute much of their success with venture capitalists (VCs) and ultimately with AOL to this quality-driven approach to architecture that was reinforced by their experiences at the SEI. “In most VC audiences,” says Carriere, “there is someone in the group whose job it is to see if you are actually going to be able to build what you say you want to build. This architecture-derived thinking was a core part of our pitch, and it seemed to resonate well with VCs. I think it served us well right up to the time when we were acquired by AOL. As we were being examined closely by AOL, we had all this thinking behind us that enabled us to answer the questions successfully.”
“In fact,” adds Woods, “AOL now says that they bought us because our technology was more flexible than anyone else's in the space.” Although flexibility was primary, Carriere and Woods, consistent with the principles of the ATAM, also looked at the tradeoffs among other key qualities. When the acquisition by AOL changed the focus, making performance and scalability more important, they knew exactly what tradeoffs they were making.
“AOL is all about scale,” says Woods. “At any given time, there are roughly two million members online at the same time, which is an enormous number. That’s a kind of scale in systems that you just don’t see anywhere else. Performance and scalability were key goals of our architecture from the beginning, even though they were secondary to flexibility. And so we were prepared to make the transitions that our acquisition by AOL required.”
Carriere and Woods retained their disciplined architecture focus throughout. “I always had the goal of being rigorous with respect to documentation of the architecture,” says Carriere. “This is a specific outgrowth of the work with Control Channel Toolkit. What I learned there was the basis for how I documented the architecture for Quack, and how I still document architecture today."
Key to Quack’s success, says Linda Northrop, Director of the SEI ATA Initiative, were the principles that Carriere and Woods applied to the way they engineered products. “They believed in architecture,” she says. “They focused on the architectural platform that generated products rather than on any particular product itself. This meant that they did not have to redesign whenever functionality changed. The architecture was designed so that functionality changes were local rather than topological; they abstracted interfaces wherever possible with external systems, and in fact, with systems they anticipated could be external. So they were able to put out revisions and change their product focus very quickly. And, it was a relatively quick task to integrate to AOL, which is a big deal for an acquisition. It was this concentration on architecture and the attention to quality attributes that differentiated them from their competitors and elevated them to a place of prominence.”
"This suggests that our architecture principles can serve any organization that is concerned about operating in the context of volatile requirements."
Northrop says that there is a lesson to be learned by the managers of large DoD systems from the success of Quack.com. “Architecture does not mean rigidity. The flexibility afforded by careful architectural design brought success to a small team of folks working in a volatile marketplace in which the definition of the project is changing on the fly,” says Northrop.
“DoD program managers often express concern about the inability to get requirements set. Imagine a world in which not only the requirements for a specific product change, but the whole product you’re building changes over and over again. That was the situation at Quack, and yet they were able to retain the same architecture. From the start, they had a strong set of architectural foundations and a reasonably accurate sense of the likely (or at least probable) change scenarios for the business. In other words, they applied the ATA philosophy. This suggests that our architecture principles can serve any organization that is concerned about operating in the context of volatile requirements. A carefully crafted architecture and disciplined architecture practices provide the necessary grounding.”
Quack.com absorbed some Netscape staff and became AOL Voice Services, which comprises a platform and any applications that are built on that platform. The AOLByPhone product is supported by the resources of the AOL Voice Services team. Woods and Carriere had leading roles on that team. Recently, however, following the completion of the version 2 platform in voice services, Woods and Carriere have moved to broader positions of impact in the AOL Technology Department. Carriere is a chief architect in the AOL Systems Infrastructure organization—a team of approximately 600 engineers responsible for building the core AOL services. Woods is VP, Systems Infrastructure. Both are now involved in a systems-wide AOL architecture effort.