Performance analysis of self-adaptive systems for requirements validation at design-time

Self-adaptation allows continuously running software systems to operate in changing and uncertain contexts while meeting their requirements in a broad range of contexts, e.g., from low to high load situations. As a consequence, requirements for self-adaptive systems are more complex than requirements for static systems as they have to explicitly address properties of the self-adaptation layer. While approaches exist in the literature to capture this new type of requirements formally, their achievement cannot be analyzed in early design phases yet. In this paper, we apply RELAX to formally specify non-functional requirements for self-adaptive systems. We then apply our model-based SimuLizar approach for a semi-automatic analysis to test whether the self-adaptation layer ensures that these non-functional requirements are met. We evaluate our approach on the design of a proof-of-concept load balancer system. As this evaluation demonstrates, we can iteratively improve our system design by improving unsatisfactory self-adaption rules.