Dynamic scheduling for real-time service with deadline constraints in green cloud data centers

Based on virtualization and DVFS (Dynamic Voltage Frequency Scaling) schemes, we study the energy-aware dynamic scheduling for real time service with deadline constraints in cloud data centers. In order to realize the tradeoff between reducing energy consumption and improving accepting new tasks, a joint scheduling strategy is proposed which combines a rolling horizon with real-time scheduling. The batch scheduling is implemented at the beginning of each planning horizon. During execution, an urgent service is deployed according to the real-time scheduling strategy. This paper first presents a mixed integer model for the batch scheduling and a heuristic algorithm for the real-time scheduling. Then a framework of the dynamic scheduling system is given to explain the basic workflow of an optimization cycle. Finally experimental results show the effectiveness of the proposed algorithm.