From executable models to deployment: A methodology for early evaluation of orchestration-based reconfiguration
Lorenzo Bacchiani, Mario Bravetti · Journal of Systems and Software · 2026
Microservice systems increasingly require sophisticated adaptation approaches to meet performance and scalability demands efficiently. However, a gap persists between the design of adaptation strategies and their reliable implementation in real-world platforms. This paper introduces an end-to-end methodology and toolchain that addresses this gap by enabling a seamless transition from high-level executable design in Behavioral Specification (ABS) to Kubernetes deployment. The methodology integrates monitoring, prediction, actuation, and orchestration synthesis into a closed-loop adaptive system. Our key contribution is Kubernetes SmartDeployer, a tool that automatically synthesises constraint-satisfying Kubernetes orchestrations from declarative specifications, using the same constraint solver as the ABS toolchain to preserve the intended architectural constraints and deployment structure specified at design time. We validate the approach through two classes of orchestration-based reconfiguration approaches: edge-cloud microservice migration and proactive-reactive architectural scaling, applied to benchmark systems and an industrial case study. Experimental results demonstrate high positive behavioral correlation (Pearson 𝑟 ≥ 0 . 8 1 , 𝑝 < 1 0 − 3 ) demonstrate strong behavioral correlation between simulated and deployed systems across multiple metrics, showing that developers can explore and preliminarily assess adaptation strategies in simulation before deployment.