Sustainability evaluation of software architectures

Long-living software systems are sustainable if they can be cost-efficiently maintained and evolved over their entire life-cycle. The quality of software architectures determines sustainability to a large extent. Scenario-based software architecture evaluation methods can support sustainability analysis, but they are still reluctantly used in practice. They are also not integrated with architecture-level metrics when evaluating implemented systems, which limits their capabilities. Existing literature reviews for architecture evaluation focus on scenario-based methods, but do not provide a critical reflection of the applicability of such methods for sustainability evaluation. Our goal is to measure the sustainability of a software architecture both during early design using scenarios and during evolution using scenarios and metrics, which is highly relevant in practice. We thus provide a systematic literature review assessing scenario-based methods for sustainability support and categorize more than 40 architecture-level metrics according to several design principles. Our review identifies a need for further empirical research, for the integration of existing methods, and for the more efficient use of formal architectural models.