Makespan Minimization for Batch Tasks in Data Centers

摘要

Makespan is one for crucial factors to determine the performance of job scheduling in cloud data center, short makespan could lead to more job throughput and less energy consumption. In this paper, we study the joint task and data assignment problem to realized makespan minimization. We propose the data migration method to overcome the memory space limitation of servers, and realize better data locality for task execution. We conduct extensive simulations, and the simulation results show that our algorithm has significant improvement on makespan reduction.