An Uncertainty-aware Evolutionary Scheduling Method for Cloud Service Provisioning

Attracted by the advantages of cloud computing, more and more services and applications are migrated to this new paradigm. As promising as it is, cloud computing also brings new challenges to many research issues, such as service scheduling. Most existing scheduling methods are offline and could not deal with the uncertainties and dynamics during the execution, especially in the dynamic cloud environment. In view of this challenge, in this paper, we propose an uncertainty-aware evolutionary scheduling method for cloud service provisioning. It aims at dealing with uncertainties during execution and updating the scheduling so as to meet the deadline and optimize the execution cost of cloud applications. Our method consists of two phases, baseline scheduling and evolutionary scheduling during execution. In baseline scheduling, we suggest a reverse-auction-based pricing mechanism for service provisioning. In evolutionary scheduling, an uncertain model with three types of uncertainties is considered and four uncertain events are discussed. Accordingly, the evolutionary scheduling policy is presented based on intermediate workflow to get a global optimal schedule, so as to improve the success rate for the execution of the cloud applications. Finally, experiments are designed and performed to demonstrate the effectiveness of our method.