An automated architectural evaluation approach based on metadata and code analysis

Abstract

Traditional methods of scenario-based software architecture evaluation rely on manual review and advanced skills from architects and developers. They are used when the system architecture has been specified, but before its implementation has begun. When the system implementation evolves, code analysis can enable the automation of this process and the reuse of architectural information. We propose an approach that introduces metadata about use case scenarios and quality attributes in the source code of the system in order to support automated architectural evaluation through of static and dynamic code analysis and produce reports about scenarios, quality attributes, code assets, and potential tradeoff points among quality attributes. Our work also describes the implementation of a code analysis tool that provides support to the approach. In addition, the paper also presents the preliminary results of its application for the architectural analysis of an enterprise web system and an e-commerce web system.