Affidavit: Automated Reliability Prediction and Analysis of Software Architectures

Authors

Abstract

Quality attributes (e.g., performance, reliability and security) are detailed in a system’s architecture and determine the fitness of purpose and satisfaction of stakeholders regarding the final product. Although research has provided methods to assess different quality attributes, few checks are automatically performed. Manually checking a quality attribute from a large and complex architecture is a time-consuming and error-prone task. This paper addresses this issue by generating stochastic models that predict and analyze reliability from a software architecture description, in an automated way. In addition, our approach has been compiled into a tool targeting system architects, the Affidavit tool. This tool is accessible from an architecture development framework and provides information about structural issues and reliability bottlenecks of systems. As a result, Affidavit allows architects to reason about the designed architecture, helping to avoid architectural arrangements that might have a negative impact on the overall system reliability, at the same time that it indicates the most suitable arrangement for specific contexts. This paper describes the tool and its implementation details, demonstrating its capabilities on practical systems.

Keywords

Software architecture, reliability, prediction, analysis

Subject

Software Engineering

Conference

INForum, September 2013