An empirical evaluation of design decision concepts in enterprise architecture

Enterprise Architecture (EA) languages describe the design of an enterprise holistically, typically linking products and services to supporting business processes and, in turn, business processes to their supporting IT systems. In earlier work, we introduced EA Anamnesis, which provides an approach and corresponding meta-model for rationalizing architectural designs. EA Anamnesis captures the motivations of design decisions in enterprise architecture, alternative designs, design criteria, observed impacts of a design decision, and more. We argued that EA Anamnesis nicely complements current architectural languages by providing the capability to learn from past decision making. In this paper, we provide a first empirical grounding for the practical usefulness of EA Anamnesis. Using a survey amongst 35 enterprise architecture practitioners, we test the perceived usefulness of EA Anamnesis concepts, and compare this to their current uptake in practice. Results indicate that while many EA Anamnesis concepts are perceived as useful, the current uptake in practice is limited to a few concepts - prominently 'rationale' and 'layer'. Our results go on and show that architects currently rationalize architectural decisions in an ad hoc manner, forgoing structured templates such as provided by EA Anamnesis. Finally, we interpret the survey results discussing for example possible reasons for the gap between perceived usefulness and uptake of architectural rationalization.