Architecture decisions: the next step. Understanding, modeling, supporting and reviewing architecture decisions

PhD ceremony: Mr. U. van Heesch, 11.00 uur, Academiegebouw, Broerstraat 5, Groningen

Dissertation: Architecture decisions: the next step. Understanding, modeling, supporting and reviewing architecture decisions

Promotor(s): prof. P. Avgeriou

Faculty: Mathematics and Natural Sciences

Software architecture is the result of a set of architecture decisions. Unfortunately, there is currently no commonly accepted approach to architecture decision modeling. Existing approaches do not satisfy all stakeholder concerns in decision description; they do not optimally support the architecting process, and they do not integrate well with the rest of the architecture documentation, which is usually arranged in multiple architectural views.

This dissertation reports on multiple empirical studies conducted to understand better the decision making process in practice. The core contribution is a framework for architecture decisions, following the conventions of the international architecture description standard ISO/IEC/IEEE 42010.

The framework consists of five interrelated viewpoints, each of which being dedicated to satisfying different stakeholder concerns in architecture decisions. The viewpoints of the framework can be used individually, or in combination, to describe the architecture decisions made in a software project.

To find out if decision viewpoints can support designers in making rational decisions, we conducted a comparative multiple case study with four groups of senior software engineering students. The results confirm that students who create decision views according to the viewpoint definition explore and evaluate candidate architectural solutions more systematically than student groups who do not use the decision framework.

Finally, this dissertation reports on a lightweight decision-centric architecture evaluation method, which uses viewpoints from the decision framework. The method uncovers and evaluates the rationale behind the most important architecture decisions made in a software project, considering all relevant forces that must be addressed by the decisions.