Developing a Framework for the Quality-Driven Migration to Microservices: A Multi-Method Design Science Study

Context: The microservices architectural style has revolutionized the way modern software systems are developed and operated. While the development of new microservices systems can leverage a wide range of resources and proven strategies, the migration of an existing monolithic system is not easily generalizable. Software architects look for guidance and predictable results in this highly individual process, in particular for generating a targeted, quality-oriented, and semi-automated decomposition.

Objective: To systematically guide software architects and developers in modernizing their software systems, we propose a holistic and quality-oriented methodology to transform monolithic applications into microservices. Our work aims to provide industry-relevant methods that address the gap between academia and practice by facilitating the transfer of knowledge.

Methods: In an overarching design science research process, we developed a framework that we implemented as a web-based application. As a preliminary work, we conducted two initial interview studies with 25 software professionals to collect evidence on the intentions, strategies, and challenges in a migration process. An essential groundwork of our framework design constitute 110 scientific publications on approaches for architectural refactoring and migration to microservices, which we reviewed over four iterations. In a multifaceted evaluation with 26 participants, we examined our methodology's capability of providing actionable guidance for practitioners. This evaluation was complemented by two longitudinal case studies in an industrial context.

Results: We provide a framework for transforming monolithic applications to microservices, along with a dedicated quality assurance concept that supports a quality-driven migration process. The evaluations among 19 software professionals showed an overall positive result in terms of effectiveness, usefulness, and usability. Two industrial case studies confirmed these promising results. Among practitioners, we discerned a need for flexibility, ease of use, and holistic guidance in a migration process. In this regard, we see potential to evolve our concept using artificial intelligence techniques for even more precise recommendations in a human-like conversational dialog.