Exploiting traceability uncertainty between software architectural models and performance analysis results

While software architecture performance analysis is a wellstudied field, it is less understood how the analysis results (i.e., mean values, variances, and/or probability distributions) trace back to the architectural model elements (i.e., software components, interactions among components, deployment nodes). Yet, understanding this traceability is critical for understanding the analysis result in context of the architecture. The goal of this paper is to automate the traceability between software architectural models and performance analysis results by investigating the uncertainty while bridging these two domains. Our approach makes use of performance antipatterns to deduce the logical consequences between the architectural elements and analysis results and automatically build a graph of traces to identify the most critical causes of performance flaws. We developed a tool that jointly considers SOftware and PErformance concepts (SoPeTraceAnalyzer), and it automatically builds model-to-results traceability links. The benefit of the tool is illustrated by means of a case study in the e-health domain.