摘要
随着网络技术与社会各行各业的深度融合与发展,现代企业数据中心面临海量数据的存储,因此需要在现有存储系统中加入新的磁盘以提升I/O并行性及容量扩展.在扩容过程中为了获得I/O负载的均衡性,需要将旧磁盘中的部分数据迁移到新加入的磁盘,然而现有的RR扩容布局算法需要迁移大量的数据,Semi-RR算法在多次扩容后存在数据布局不均匀性等问题.本文提出了一种新的RAID4容量扩展算法USR4,该算法在扩容过程中体现出均匀数据布局性、最小化数据迁移量和快速扩容等特性.相比于RR扩容算法,性能评价结果表明,USR4算法减少了80%-88.9%的数据迁移量,缩短了40%-62.23%总扩容时间.
With the development and deeply integration of network technology and Community businesses,modern data centers of enterprise are facing with the need for storing massive data. Thus,disk additions to an existing storage system can increase the I/O parallelism and expand the storage capacity simultaneously. To regain I/O balancing in the process of RAID scaling,RAID scaling requires moving certain data blocks onto newly added disks. However,existing RR scaling algorithm require migrating all the data,and SemiRR scaling algorithm can't make data uniform distribution after multiple scaling. The paper proposed a new RAID4 scaling algorithm USR4,which delivers the characteristics of uniform data layout,minimal data migration amount and fast scaling. Compared with the Round-Robin scaling algorithm,the result of mathematicanalysis evaluation illustrates that USR4 algorithm decreased the data migration amount by 80%-88. 9%,shorten the total scaling time by 40%-62. 23%.
引文
[1]Gonzalez J L,Cortes T.Increasing the capacity of RAID5 by online gradual assimilation[C].International Workshop on Storage Netw ork Architecture and Parallel I/os(SNAPI'04),ACM,2004:17-24.
[2]Goel A,Shahabi C,Yao S Y D,et al.SCADDAR:an efficient randomized technique to reorganize continuous media blocks[J].Journal in Computer Virology,2002,2(1):473-482.
[3]Zhang G,Shu J,Xue W,et al.SLAS:An efficient approach to scaling round-robin striped volumes[J].Acm Transactions on Storage(TOS),2007,3(1):1-39.
[4]Zhang G,Zheng W,Shu J.ALV:a new data redistribution approach to RAID-5 scaling[J].IEEE Transactions on Computers(TOC),2010,59(3):345-357.
[5]Wu C,He X.GSR:a global stripe-based redistribution approach to accelerate RAID-5 scaling[C].International Conference on Parallel Processing(ICPP'12),IEEE,2012:460-469.
[6]Zhang G,Li K,Wang J,et al.Accelerate RDP RAID-6 scaling by reducing disk I/Os and XOR operations[J].IEEE Transactions on Computers(TOC),2014,64(1):32-44.
[7]Zhang G,Wu G,Lu Y,et al.Xscale:online X-Code RAID-6 scaling using lightw eight data reorganization[J].IEEE Transactions on Parallel&Distributed Systems(TPDS),2016,27(12):3687-3700.
[8]Patterson D A,Gibson G A,Katz R H.A case for redundant arrays of inexpensive disks(RAID)[M].University of California at Berkeley,1987.
[9]Jin H,Zhou X,Feng D,et al.Improving partial stripe write performance in RAID level 5[C].The Second IEEE International Caracas Conference on Devices,Circuits and Systems(ICCDCS'02),IEEE,2002:396-400.
[10]Mao Yu.Research on capacity promotion technology of disk array and disk in block level[D].Wuhan:Huazhong University of Science&Technology,2014.
[11]Li Jun-yao.Research of a RAID-5 system expansion technique[D].Wuhan:Huazhong University of Science&Technology,2013.
[12]Feng Dan,Peng Li.Novel scheme for online capacity expansion in RAID system[J].Computer Application and Research,2006,23(12):244-246.
[13]Gao Shuai.Research on rotated based scaling scheme for RAID6storage system[D].Hefei:Univiersity of Science and Technology of China,2015.
[10]毛宇.磁盘阵列及磁盘块级容量提升技术研究[D].武汉:华中科技大学,2014.
[11]李俊尧.一种RAID-5快速扩容技术的研究[D].武汉:华中科技大学,2013.
[12]冯丹,彭丽.一种新颖的RAID系统在线扩容方案[J].计算机应用研究,2006,23(12):244-246.
[13]高帅.基于轮转部署的RAID6存储系统扩容问题研究[D].合肥:中国科学技术大学,2015.