Research On Task Scheduling Strategy For Green Data Center Partially Powered By The Renewable Energy

With the improvement of environmental awareness and the approaching of energy crisis,new green data centers partially powered by the renewable energy appear on the basis of the traditional data center.The task scheduling strategy is undoubtedly an important research focus in data center energy management.The task scheduling strategy of the new green data center should not only consider the tasks to be completed in the shortest time,but also need to consider the utilization rate of renewable energy and reduce the carbon emission.How to schedule the task in the green data center is a key factor which affects the successful implement of the data center.Therefore,this studying is meaningful for both of theory and practicality.This thesis conducted a series of research and exploration on the construction of new green data center system model and the design and verification of task scheduling strategy,including:First of all,according to the current situation,a green data center system model is designed.Directed Acyclic Graph(DAG)model is adopted to describe the dependent tasks,and considers the heterogeneous factors of data center as well.Secondly,a Smart Strategy based on Green-energy & Price(SSGP)is proposed.SSGP ensures that under the circumstance of task dependencies are not destroyed,more reasonable computing nodes and time slots are selected according to the characteristics of the task,which also increases the utilization of renewable energy,reduces operating costs and reduces the task completion time.The experimental results show that the SSGP is an efficient task scheduling method compared with the completely strategy based on green-energy(CSG),the completely strategy based on price(CSP),and the strategy based on green-energy & price(SGP).Finally,there is further optimization to the proposed scheduling strategy.Facing the problem of the scheduling sequence of the same layer tasks based solely on time factors into considerationand too one dimensional,introduce static list scheduling for more reasonable task priorities.The experimental results show that SSGP is improved.