Safe Dynamic Reconfiguration of Concurrent Component-based Applications

International audienceCloud computing and cyber-physical systems involve software capable of adapting at run time to remain compliant with user demands and environmental constraints. This calls for extending the life cycle of software systems with a reconfiguration step to go beyond analysis, design, development and deployment. Existing approaches compute a new valid configuration at design time, at run time, or both, inducing computational or validation overheads for each reconfiguration step. We propose an approach that relies on variability models to acquire a representation of the set of valid configurations of a system. We use feature models to automatically generate a JavaBIP run-time variability model. The generated model monitors and controls the application behaviour by intercepting reconfiguration requests and executing them in such a manner as to ensure that all reachable configurations are valid without the need of pre-computing the possible configurations neither at design time nor at run-time while only inducing a minimal run-time computational overhead