一种云环境中能量高效的任务分配方法

In this paper we propose a general algorithm ReBaSCP to energy aware allocate data-intensive tasks by powering down partial nodes in a deployed mapreduce cluster without modifying the replication strategy. ReBaSCP, which is a two-step method, selects minimal nodes to process the data-intensive task at first to coarsely reduce energy of the cluster, and then balances the workload of the selected nodes to further reduce the energy consumption of the selected nodes in a finer granularity. Experimental results show that our solution outperforms the existing methods..