对等结构的混合集群的层次化存储策略研究  

RESEARCH ON HIERARCHICAL STORAGE STRATEGY ON P2P HYBRID CLUSTER

作　　者：涂超凡 张启飞[1] 陈应庄[1] 张昱昶 刘二腾[1] 干红华[1] 张尉东[2] 白凡[3] 郑贤榕 

机构地区：[1]浙江大学软件学院,浙江宁波315048 [2]北京大学电子工程和计算机科学系,北京100871 [3]西北大学电子工程和计算机科学系,美国埃文斯顿60201 [4]浙大网新中央研究院,浙江杭州310030

出　　处：《计算机应用与软件》2017年第4期6-10,69,共6页Computer Applications and Software

基　　金：国家关键科技支撑项目(2014BAH23F03);国家环境公益项目(2013A610064);宁波自然科学基金项目(2013B10036);宁波智能交互多媒体远程教育系统项目(2013B10036)

摘　　要：随着分布式存储系统的规模迅猛增长,能耗问题日益凸显。在存储集群混合低能耗节点已成为解决能耗问题的重要途径之一。针对如何在对等结构的混合集群中区别使用不同类型存储节点的问题,在一个对等结构分布式系统ZDFS上设计并实现了一种基于对等结构的层次化存储策略——虚拟节点分层重映射(VHR)。该策略不影响对等结构的高拓展特性,使系统可自动区分使用不同类型存储节点,可操作性强。实验在真实X86和ARM混合集群上进行。实验表明,VHR实验组运行良好,充分发挥了不同类型存储节点的优势,在性能降低不明显的情况下,整体能耗水平降低了44.8%。With the scale of distributed storage systems growing rapidly, the problem of energy consumption has become increasingly prominent, and storage cluster hybrid low-power node has become one of the important ways to solve this problem. To face the problem of how to use different types of storage nodes discriminatively, in this paper, a hierarchical storage strategy called vnode hierarchical remapping （VHR） is designed and implemented based on a P2P distributed system of ZDFS. The strategy didn’t affect the high scalability of P2P, so that the system can automatically distinguish between the use of different types of storage nodes with strong operability. Experiments are carried out on a real X86 and ARM hybrid cluster, and experiments show that the VHR works well and plays to the advantages of different types of storage nodes. Under the condition of the performance degradation is not obvious, the overall level of energy consumption was reduced by 44.8%.

关 键 词：存储系统 对等结构 混合集群 层次化 

分 类 号：TP3[自动化与计算机技术—计算机科学与技术]