Towards the predictive analysis of cloud systems with e-Motions

Abstract

Current methods for the predictive analysis of software systems are not directly applicable on self-adaptive systems as cloud systems, mainly due to their complexity and dynamism. To tackle the difficulties to handle the dynamic changes in the systems and their environments, we propose using graph transformation to define an adaptive com- ponent model and analysis tools for it, what allows us to carry on such analyses on dynamic architectures. Specifically, we use the e-Motions system to define the Palladio component model, and simulation-based analysis tools for it. Adaptation mechanisms are then specified as generic adaptation rules. This setting will allow us to study different mechanisms for the management of dynamic systems and their adaptation mechanisms, and different QoS metrics to be considered in a dynamic environment.