Data Management in Microservices: State of the Practice, Challenges, and Research Directions

Abstract: We are recently witnessing an increased adoption of microservice architectures by the industry for achieving scalability by functional decomposition, fault-tolerance by deployment of small and independent services, and polyglot persistence by the adoption of different database technologies specific to the needs of each service. Despite the accelerating industrial adoption and the extensive research on microservices, there is a lack of thorough investigation on the state of the practice and the major challenges faced by practitioners with regard to data management. To bridge this gap, this paper presents a detailed investigation of data management in microservices. Our exploratory study is based on the following methodology: we conducted a systematic literature review of articles reporting the adoption of microservices in industry, where more than 300 articles were filtered down to 11 representative studies; we analyzed a set of 9 popular open-source microservice-based applications, selected out of more than 20 open-source projects; furthermore, to strengthen our evidence, we conducted an online survey that we then used to cross-validate the findings of the previous steps with the perceptions and experiences of over 120 practitioners and researchers. Through this process, we were able to categorize the state of practice and reveal several principled challenges that cannot be solved by software engineering practices, but rather need system-level support to alleviate the burden of practitioners. Based on the observations we also identified a series of research directions to achieve this goal. Fundamentally, novel database systems and data management tools that support isolation for microservices, which include fault isolation, performance isolation, data ownership, and independent schema evolution across microservices must be built to address the needs of this growing architectural style.