Evidence in software architecture, a systematic literature review

Abstract

Context: Software architecture (SA) plays pivotal role in development and maintenance of large software systems. Architectural decisions impact all subsequent phases in software development life cycle. Structure of large software systems has been in discussion since early 70s but software architecture started emerging as separate discipline since mid-90s. Objective: Researchers have been publishing empirical studies (i.e. case studies, experiments, experience reports) in SA discipline. Empirical literature has been aggregated in other disciplines of software engineering but no such effort has been made in SA. Objective of this study is to aggregate and synthesize the empirical evidence from literature of software architecture to report the trends, patterns and knowledge gaps. Method: To synthesize the empirical work in SA, we conducted a systematic literature review (SLR). This study reports the results of SLR based on 247 included primary studies. Results: Based on the percentage of 247 primary studies we identified that SA Evaluation (28%), non- functional requirements related work (22%) and SA Design (12%) are relatively mature sub areas of SA. Most of the empirical work (59%) in SA uses case study research method. Experiments (20%) and experience reports (14%) are also employed in empirical SA work. Conclusion: SA discipline is maturing as few sub areas of SA are mature and some areas are new and being developed. We have also identified few emerging trends in SA i.e. service oriented architecture, Product line architecture, Aspect oriented architecture, and model driven architecture.