How does PrEstoCloud address the constraint of drastically changing workload scenarios?
Insufficient cloud-based infrastructures, which is called under-provisioning, allocated to the running application may unavoidably harm the performance and result in Service Level Agreement (SLA) violations, while over-provisioning may also cause cost waste and resource idleness. This constraint has been a challenging issue due to runtime variations in workload density especially for drastically changing workload scenarios. To address this constraint, the “Workload Predictor” component defined in the PrEstoCloud architecture is able to predict the dynamic workload of the underlying cloud infrastructure at run-time, and hence enable refinement process for quality assurance. Moreover, the Resources Adaptation Recommender (RARecom) is able to exploit these predictions along with real-time monitoring data to enact proper cloud application adaptations, which are constantly improved through a dedicated feedback mechanism.