电子学报 ›› 2019, Vol. 47 ›› Issue (11): 2420-2431.DOI: 10.3969/j.issn.0372-2112.2019.11.026

• 综述评论 • 上一篇    下一篇

RAID系统扩容方案研究综述

元铸, 谢平, 耿生玲   

  1. 青海师范大学计算机学院, 青海西宁 810008
  • 收稿日期:2019-02-01 修回日期:2019-07-25 出版日期:2019-11-25 发布日期:2019-11-25
  • 通讯作者: 谢平
  • 作者简介:元铸 男,1993年出生,硕士研究生.主要研究方向是海量存储系统、新型存储技术.E-mail:761443043@qq.com;耿生玲 女,1970年出生,博士,教授,CCF会员(E20033713M).主要研究方向为计算理论、数据挖掘、控制与决策的研究.E-mail:geng_sl@126.com
  • 基金资助:
    国家自然科学基金(No.61762075,No.61862055);青海省自然科学基金(面向大数据环境的磁盘阵列高效扩容架构研究);赛尔网络下一代互联网技术创新项目(No.NGII20180116);青海省物联网重点实验室建设专项(No.2017-ZJ-Y21)

Summary of Research for RAID System Scaling Schemes

YUAN Zhu, XIE Ping, GENG Sheng-ling   

  1. The Computer College of Qinghai Normal University, Xining, Qinghai 810008, China
  • Received:2019-02-01 Revised:2019-07-25 Online:2019-11-25 Published:2019-11-25

摘要: 随着云计算、物联网和人工智能等IT技术驱动数字经济产业的繁荣发展,现代企业信息化需求对数据中心存储能力提出了更高的要求和挑战.RAID系统因具备良好的数据存储可靠性和磁盘阵列可扩展性而得到广泛应用.为了满足海量数据对存储容量日益增长的需求,业界普遍采用扩容现有RAID系统以应对海量数据的存储问题.电子商务、Web服务和金融等行业对数据的实时访问,使得数据中心必须为用户提供7*24的高质服务响应,然而数据迁移量,负载均衡和扩容开销等因素都会影响扩容的效率,因此如何设计出一种高效的扩容方案越来越受到科研人员的关注.本文根据研究对象的不同将RAID扩容方案分为:基于块存储、对象存储、文件系统存储的扩容方案,同时根据RAID扩容方案研究历程和优化策略的不同,又可分为优化数据迁移过程和减少数据迁移量的扩容方案.文章结合不同的应用场景介绍了典型和常见的扩容方案,并从评价扩容方案性能各项重要指标的角度详细分析了现有的RAID扩容方案,并指出各种扩容方案的不足以及未来可能的发展方向.如何权衡影响扩容方案性能的各项因素,设计出数据迁移少,负载均衡好,扩容开销低的高效扩容方案,将会是今后相当长一段时间海量数据存储研究的热点问题.

关键词: 海量数据存储, RAID系统扩容, 扩容方案, 扩容方案性能指标, 高效扩容

Abstract: With the development of digital economy industry driven by IT technologies such as cloud computing,internet of things and artificial intelligence,modern enterprise information demands put forward higher requirements and challenges for data center storage capacity.Due to data storage reliability and redundant arrays of independent disks scalability,RAID system is widely used.In order to fulfill the requirements of massive data for the increasing storage capacity,the industry generally adopts a method of scaling RAID system to solve the problem of storing massive data.E-commerce,web service and finance access data on real-time,which make the data center must provide high-quality service response for users with 7*24,but the factors of data migration,load balane and scaling cost will affect the efficiency of RAID scaling.Therefore,how to design a fast and efficient scaling scheme is getting more and more attention from researchers.According to the different investigative object,this paper classifies RAID scaling approaches into basing on block,object and file system.Meanwhile,according to the investigative developing process and different optimization strategies of RAID scaling,the approaches also can divide into optimizing data migration process and reducing the number of data to be moved.From the perspective of evaluating the performance of the scaling schemes,this article introduces the typical and common scaling schemes by different application scenarios that points out the shortcomings of various scaling schemes and possible improvements.How to balance the factors that affect the performance of the scaling schemes and design a high-performance scaling scheme with less data migration,good load balance,and low scaling cost,which will be a hot issue for massive data storage research for a long time in the future.

Key words: massive data storage, RAID scaling, scaling schemes, index of scaling performance, efficient scaling

中图分类号: