基于随机森林的能耗预测调度  

Predictive Scheduling of Energy Consumption Based on Random Forest

作　　者：许晔炫 潘景山 王继彬 XU Ye-xuan;PAN Jing-shan;WANG Ji-bin(School of Computer Science and Technology,Shandong Jianzhu University;National Supercomputer Center in Jinan,Ji'nan 250101,China)

机构地区：[1]山东建筑大学计算机科学与技术学院 [2]国家超级计算机济南中心,山东济南250101

出　　处：《软件导刊》2022年第5期89-95,共7页Software Guide

基　　金：山东省自然科学基金项目(ZR2019QF014)。

摘　　要：云计算资源池由大量同构物理机构成,但由于用户和应用产生的需求不同,导致每个节点产生的计算能耗不同,使得某些节点长期处于高能耗状态,出现热点现象,而节点长期处于热点会损害物理机使用寿命。针对以上问题,提出一种基于随机森林的能耗预测调度算法(ECPRF),根据能耗预测进行虚拟机迁移,平衡每台物理机的能耗。实验表明,该调度算法可以有效平衡服务器能耗,避免出现热点现象。The cloud computing resource pool is composed of a large number of homogeneous physical machines.However,due to the different demands generated by users and applications,the computing energy consumption generated by each node is different,which makes certain nodes have long term high energy consumption and hot spots.Long-term hot spots on the node will damage the life of the physical machine.To solve the above problems,this paper proposes an energy prediction scheduling algorithm based on random forest(ECPRF),which performs virtual machine migration according to energy consumption prediction,and balances the energy consumption of each physical machine.Experiments show that the scheduling algorithm can effectively balance server energy consumption and avoid hot spots.

关 键 词：云计算 随机森林 能耗预测 能耗预测调度算法 

分 类 号：TP312[自动化与计算机技术—计算机软件与理论]