动态盘阵D/H分布与基于控制理论的在线重构

【网学提醒】：本文主要为网上学习者提供动态盘阵D/H分布与基于控制理论的在线重构，希望对需要动态盘阵D/H分布与基于控制理论的在线重构网友有所帮助，学习一下吧!

资料包括： 论文(11页11272字) 
说明：

摘 要：由于能够提供高性能I/O,盘阵被广泛采用.但以往的盘阵扩展性不足.而用户或应用程序对外存容量和I/O性能需求是变化的,盘阵系统本身必须有很强的扩展性,以适应系统的I/O需求.因此,由于既具有盘阵的高性能I/O,又能通过增加或减去设备后进行数据重构实现性能的扩展,动态盘阵具有广泛的前景.动态盘阵的技术热点是数据分布算法和在线自适应数据重构技术,使得盘阵的性能和容量能够随着系统的扩展而伸缩,同时使得盘阵动态扩展时的数据重构对系统的影响非常小.主要工作是:第1,对动态盘阵的数据分布展开研究,并提出一种新的数据分布算法(D/H分布).在D/H分布中,盘阵扩展时始终保持各设备上空间和负载的平衡性,同时扩展时重构的数据最少;第2,针对D/H分布,提出基于控制理论的数据重构技术,使得盘阵动态扩展时的在线数据重构对请求QoS的影响非常小,同时使得数据重构能够尽快完成;第3,研究中针对Sprite trace和合成负载进行了大量模拟实验,结果表明,提出的基于控制理论的数据重构技术行之有效.

关键词: 盘阵;动态扩展;动态盘阵;数据分布;算法;在线重构;控制理论

D/H Placement and On-Line Data Reorganization Based on Control Theory in Dynamic Disk Array

Abstract：Disk Array is adopted widely because of its high performance I/O. To adapt the need of applications’ changeable I/O performance, I/O storage subsystem should be highly scalable. So DDA (dynamic disk array), which can scale adaptively, is an ideal system. The key technology of DDA is its data placement algorithm and online data reorganization algorithm. The main contribution of the paper is: first, a detailed study on DDA data placement is conducted and a new placement method, D/H, is presented. In D/H placement, the space in DDA is balanced after scale, and the reorganization cost is minimized; then, an online data reorganization algorithm based on control feedback theory is provided. With this strategy, the reorganization in DDA does little impression to the system QoS, and under this condition, data reorganization can be accomplished as quickly as possible; finally, simulation results show that Online Data Reorganization based on Control Theory is useful.

Key words: disk array; scaling dynamically; dynamic disk array; data placement; algorithm; on-line reorganization; control theory

随着CPU和内存等微电子部件性能的迅猛提高,磁盘I/O成为计算机系统的主要瓶颈之一.因此作为一种高性能的I/O子系统,盘阵被广泛应用.
但是以往的盘阵大都是一种静态的,也就是说,其规模和容量、性能在系统构建之初就已经确定,不能随需求的变化而动态扩展.而在实际的需求中,一方面,不同的应用,I/O需求不同.比如有的科学计算应用I/O需求巨量,而有的应用I/O需求低一些;有的情况下,即使同种类型的应用程序,在不同的时间段运行,其I/O需求差别达到一个数量级;另一方面,有的盘阵系统本身规模也是变化的.比如在集群分布式系统中,在将各结点的外设构成虚拟盘阵的I/O子系统时,由于分布式环境是一种变化的系统,经常有结点的加入或退出[2,3],从而导致虚拟盘阵规模的变化,因此,为了满足这些应用程序的需要或适应系统环境的变化,盘阵应该是可动态扩展的系统,也就是动态盘阵.

目录：
1 相关工作
2 D/H分布
3 基于控制理论的在线重构
4 模拟实验
5 结束语
参考文献：
Shen XH, Choudhary AN, Matarazzo C, Sinha P. A distributed multi-storage resource architecture and I/O performance prediction for scientific computing. Cluster Computing, 2003,6(3):189200.
Harting H. An I/O architecture for microkernel-based operationg systems. Technical Report TUD-FI03-08-Juli-2003, Dresden, TU Dresden, 2003.
Ho TK, Lee JYB. A row-permutated data reorganization algorithm for growing server-less video-on-demand systems. In: Lee S, Sekiguchi S, eds. Proc. of the 3rd IEEE/ACM Int’l Symp. on Cluster Computing and the Grid (CCGRID 2003). Tokyo: IEEE/ACM Press, 2003. 4451.
Liu J, Tang YH, Wang L, Jiang YH, Yang XJ. The dynamic mass storage system. Journal of Computer Research and Development, 2002,39(Suppl.):162167 (in Chinese with English abstract).
Ghandeharizadeh S, Kim D. On-Line reorganization of data in scalable continuous media servers. In: Feitelson DG, Rudolph L, eds. Proc. of the 7th Int’l Conf. on Database and Expert Systems Applications. Lecture Notes in Computer Science, Zurich, 1996. 751768.
Goel A, Shahabi C, Yao S-YD, Zimmermann R. SCADDAR: An efficient randomized technique to reorganize continuous media blocks. In: Chaudhuri S, Carey M, Garcia-Molina H, eds. Proc. of the 18th Int’l Conf. on Data Engineering (ICDE 2002). San Jose: IEEE CS Press, 2002. 473482.
Lu CY, Alvarez GA, Wilkes J. Aqueduct: Online data migration with performance guarantees. In: Chase J, Cole J, eds. Proc. of the USENIX Conf. on File and Storage Technologies (FAST). Monterey, 2002. 219230.
Abdelzaher TF. An automated profiling subsystem for QoS-aware services. In: Sztipanovits J, Abdelzaher TF, Atkins EM, eds. Proc. of the 6th IEEE Real Time Technology and Applications Symp. Washington DC: IEEE Press, 2000. 208217.
Lumb CR, Merchant A, Alvarez GA. Facade: Virtual storage devices with performance guarantees. In: Chase J, Cole J, eds. Proc. of the 2nd USENIX Conf. on File and Storage Technologies (FAST). San Francisco, 2003. 131144.
Madell T, Madell T. Disk and File Management Tasks in HP-UX. Printice-Hall: Pearson Education POD, 1997.
Franklin GF, Powell DJ, Workman ML. Digital Control of Dynamic Systems. 3rd ed., New York: Addison-Wesley, 1998.
Li BC, Nahrstedt K. A control-based middleware framework for quality of service adaptations. IEEE Journal of Selected Areas in Communication, Special Issue on Service Enabling Platforms, 1999,17(9):16321650.
Lu CY, Stankovic JA, Tao G, Son SH. Feedback control real-time scheduling: Framework, modeling, and algorithms. Journal of Real-Time Systems, 2002,22(1/2):85126.
Holland M, Gibson GA. Parity declustering for continuous operation in redundant disk arrays. In: Mukherjee S, McKinley KS, eds. Proc. of the 5th Int’l Conf. on Architectural Support for Programming Languages and Operationg Systems (aspLOS-V). Boston MA: ACM Press, 1992. 2335.
[15] Gribble S, Manku G, Roselli E, Brewer E. Self-Similarity in file systems. In: Gibson TJ, Gribble SD, eds. Proc. of the 1998 ACM SIGMETRICS Joint Int’l Conf. on Measurement and Modeling of Computer Systems. Madison: ACM Press, 1998. 141150.
[16] Jain R. The Art of Computer Systems Performance Analysis. New York: John Wiley & Sons, Inc. 1991.
[17] Smith W, Foster I, Taylor V. Predicting application run time using historical information. In: Feitelson DG, Rudolph L, eds. Proc. of the Workshop on Job Scheduling Strategies for Parallel Processing. Lecture Notes in Computer Science, San Juan, 1999. 202219.
[18] Schindler J, Griffin JL, Lumb CR, Ganger GR. Track-Aligned extents: Matching access patterns to disk drive characteristics. In: Chase J, Cole J, eds. Proc. of the Conf. on File and Storage Technologies (FAST). Monterey, 2002. 259274.
[19] Lee EK, Katz RH. An analytic performance model of disk arrays. In: Lazowska ED, Felten EW, eds. Proc. of the 1993 ACM SIGMETRICS Conf. on Measurement and Modeling of Computer Systems. Santa Clara: ACM Press, 1993. 98109.
[20] Baker MG, Hartman JH, Kupfer MD, Shirriff KW, Ousterhout JK. Measurements of a distributed file system. In: Proc. of the 13th ACM Symp. on Operating System Principles. Pacific Grove: ACM Press, 1991. 198213.

作者点评：
由于能适应系统变化的I/O需求和系统规模的可扩展性需求,动态盘阵有着广泛的应用.但是动态盘阵为了实现其性能的扩展,必须选择很好的数据分布算法和在线重构策略,使得盘阵扩展后设备上空间保持一种平衡性,并使得动态扩展时引起的数据迁移对当前系统的QoS影响较小.为此,本研究提出了D/H数据分布算法和基于控制理论的在线重构策略.研究中,D/H扩展与数据分布算法能使扩展开销、空间平衡性和负载平衡性达到优化.基于D/H数据分布,对这种在线重构策略进行了模拟,实验表明,基于控制理论的在线重构策略,能使系统QoS和动态重构速度达到较优.
但是基于控制理论的算法关键是基于当前输入进行反馈,预测后面的情况.在本研究中采用线性回归模型.实验表明,该模型太简单,与实际的客户端负载情况有误差,因而不能得到最优的结果.下一步工作需要寻找更准确的反馈函数,使重构达到最优.