Scaling Architecture Evaluations Within Real-World Constraints

This presentation was created for the SATURN conference series and does not necessarily reflect the positions and views of the Software Engineering Institute.

Many elements of formal software architecture evaluation approaches (e.g., ATAM and SAAM) have found their way into industrial practice–for example, using quality scenarios for writing high-quality requirements and prioritizing quality attributes using the utility tree. However, commercial industry has been slow to adopt these methods in their entirety because of perceived cost and effort. We have repeatedly witnessed real-world situations in which industrial architects facing tight schedule time boxes or budget constraints are precluded from using these methods. As a result, they often fall back to informal “evaluations” based on their judgment and experience. Even though they may use a few elements from formal methods, these lightweight evaluations are generally undertaken in an ad hoc manner.

We have developed a framework providing industry-friendly structure and guidance for tailoring lightweight architectural evaluations that work within common business constraints. This presentation describes the methodology and its pros and cons, and concludes with the results and lessons learned from the industrial pilot in which it was applied.