Coordinated run-time adaptation of variability-intensive systems: an application in cloud computing

Distributed systems, such as cloud systems or cyber-physical systems, involve the orchestration of different variability-intensive, adaptive sub-systems. Each of these sub-systems may perform adaptations simultaneously and independently from each other. Yet, if dependencies between the adaptations of the sub-systems are not considered, this may lead to conflicting adaptations or untapped synergies among adaptations. This paper introduces FCORE, a model-based approach, which facilitates coordinating adaptations among variability-intensive systems. The permissible run-time reconfigurations of each system is specified by an FCORE model, which combines feature models used in Dynamic Software Product Lines with goal models. FCORE models are mapped to constraint satisfaction problems to determine conflicts and synergies among the adaptations of the systems during execution. We demonstrate the FCORE approach by using a cloud system as a typical exemplar for a distributed system. The cloud system is part of an industrial use case concerned with offering value-added cloud services.