Lessons Learned in Model-Based Reverse Engineering of Large Legacy Systems

Abstract

Large technologies companies that offer software modernization and maintenance services for legacy software applications in diverse sectors such as banking, insurance, healthcare and public sector, face a significant challenge. Legacy systems were usually developed in old programming languages, often have outdated documentation and the processes used for software development were immature. Modernization and maintenance projects include tasks such as source code analysis with high effort and time costs, and an important risk of misunderstanding. In the literature, model-driven reverse engineering (MDRE) approaches promise to address these challenges successfully, but most of existing proposals are focused on a concrete technological stack. This paper aims to present the preliminary results and lessons learned when adopting MDRE in a large multinational company, providing a series of reflections and open issues to reduce the gap between academia and industry. It introduces STRATO, a corporate solution that proposes a MDRE approach focused on a high flexibility to incorporate new programming languages. It reads source code and through model-to-model transformations convert it into platform independent conceptual, persistence and business logic models. Preliminary outcomes, lessons learned and open issues concerning MDRE industry adoption are presented.