Simulation Approaches for Supporting Microservice Architectures: A Systematic Review

<p>  </p> <p>Microservice Architecture (MSA) has become a dominant architectural style due to its scalability, modularity, and resilience. However, evaluating and optimising MSA systems remains complex, particularly in performance assessment, architectural decision-making, and system behaviour analysis. Simulation has been explored as a technique to address these challenges, yet a comprehensive synthesis of its applications, methodologies, and effectiveness in MSA research is lacking. This Systematic Literature Review (SLR) investigates how simulation has been applied to MSA research, addressing 11 research questions, covering aspects such as the motivations for using simulation, architectural decisions it supports, quality attributes analysed, modelling techniques used, validation methods, and available simulation tools. The study further examines the adoption and adaptation of simulation tools across academia and industry. Following Kitchenham’s SLR guidelines, a systematic search was conducted across four major digital libraries. The search initially retrieved 1,434 studies, which were refined through rigorous inclusion and exclusion criteria. To enhance coverage, a snowballing approach was applied for additional validation. 68 primary studies were selected for inclusion and data extraction and analysis to respond to our 11 research questions. The findings highlight key aspects of simulation in MSA. Simulation aids architectural decisions by modelling performance, scalability, resilience, and deployment strategies. Commonly analysed quality attributes include fault tolerance, load balancing, and resource optimisation.. Validation methods remain inconsistent, and real-world datasets are rarely integrated. Simulation tools are diverse in their capabilities, with varying levels of adoption and adaptation in academia and industry. This SLR provides a structured synthesis of simulation applications in MSA, identifying trends, challenges, and research gaps. The findings reveal persistent gaps in areas such as migration support, security and fault propagation modelling, and real-world validation. They highlight the need for more realistic and adaptive simulation frameworks that integrate authentic artefacts, support runtime calibration, and explore emerging opportunities such as LLM-assisted modelling. These directions establish an agenda for advancing simulation as a proactive capability in microservice system design and resiliency.</p>