Energy-Aware Tasks Scheduling with Deadline-constrained in Clouds

As the expanding of cloud datacenter and the soaring demand for various types of services and resources, the energy consumption problem is continuously mushrooming. Unfortunately, existing scheduling algorithms mainly pursue timeliness, but take little account the energy factor. On the other hand, Service Level Agreement (SLA) plays a key role in guaranteeing the QoS for users. In SLA, deadline (response time) is one of the QoS metrics that users are mostly concerned about. However, the deadline set in SLA is likely to be violated when there are real-time tasks with different tight deadline-constrained. Therefore, it is a valuable research issue to propose a scheduling algorithm which can save energy while meeting the deadline. In view of this challenge, an energy-aware tasks scheduling method with deadline-constrained is proposed in this paper by exploiting the computing parallelism of divisible task. In our method, the urgency level is defined to determine the scheduling order of the real-time tasks waiting to be processed. Accordingly, two energy-aware tasks scheduling algorithms are proposed by considering whether the load of task is divisible. Finally, experimental evaluations are presented for validating the method.