TQPropRefiner: Interactive Comprehension and Refinement of Specifications on Transient Software Quality Properties

Microservice-based systems are exposed to transient behavior caused, for example, by (frequent) deployments, failures, or self-adaption. The potential complexity of transient behavior scenarios makes specifying flawless transient behavior requirements challenging. Still, the required approaches and tooling to comprehend transient behavior and refine the requirements are lacking. This paper aims to address this gap by providing a structured interactive approach that assists software architects in comprehending transient behavior and refining requirements. The prototypically implemented TQPropRefiner allows specifying transient behavior requirements using PSP. Then, TQPropRefiner uses runtime verification to evaluate requirement satisfaction on time-series data, e.g., from Chaos Experiments. TQPropRefiner visualizes the system’s behavior and requirement satisfaction to foster comprehension. Based on the gathered insights, users can refine their requirements. In particular, TQPropRefiner currently supports refining timing constraints and simple predicates. Finally, we evaluated the feasibility and practical applicability of our early approach in a qualitative user study with five industry experts. All participants could interpret the results, and four solved the refinement task successfully. Despite currently limited support of PSP and refinement strategies, the preliminary results indicate that the approach can facilitate understanding transient behavior requirements among software architects and assist in the refinement process. Thus, our work is a first step toward facilitating the comprehension of transient behavior and refinement of requirements.