期刊文献+

具有磨损均衡意识的混合固态硬盘FTL算法 被引量:5

Wear Leveling Aware FTL for Hybrid Solid State Disks
在线阅读 下载PDF
导出
摘要 在采用SLC(Single-Level-Cell)和MLC(Multi-Level-Cell)闪存的混合固态硬盘设计中,SLC和MLC之间的写数据分配和磨损均衡是混合固态硬盘闪存转换层设计的关键问题之一.针对此问题,提出一种具有磨损均衡意识的混合固态硬盘闪存转换层算法——WLAFTL(Wear Leveling Aware Flash Translation Layer).首先,它提出了一种动态的基于磨损均衡思想和请求大小融合的数据分配机制,即根据SLC和MLC的磨损速率来动态调整热数据识别阈值的大小,然后将小的写请求分配到SLC、大的写请求分配到MLC.其次,它提出了一种基于磨损均衡思想与数据先进先出(FIFO)调度策略融合的SLC冷数据回收/迁移机制,减少由SLC向MLC迁移的数据量.实验结果显示,与ComboFTL和CFTL算法相比,在使用相同地址映射机制的条件下,WLAFTL算法的平均响应时间分别平均有13.6%和12.7%的改善,总的擦除次数分别平均减少9.2%和20.4%,同时能够更好地实现SLC和MLC之间的磨损均衡. With the rapid development of NAND flash memory technologies,replacing hard disks with NAND flash based solid state disks(SSDs)has become a recently emerging trend in computer storage domain.Nowadays,SSDs are mainly based on homogeneous NAND flash chips such as single-level cell(SLC)flash chips and multi-level cell(MLC)flash chips.SLC and MLC chips are distinctive in terms of capacity,performance,and endurance.Specifically,SLC has the advantage of fast read/write,low power consumption and long life,and MLC has the advantage of capacity units with low cost.Therefore,it is a promising method to design hybrid SSDs,consisting of both SLC and MLC flash chips,to achieve a response time and lifetime as same as a SLC-based SSD while maintaining the price of a MLC-based SSD.In order to realize these design goals of hybrid SSDs,the key issues are:(1)how to allocate data into SLC and MLC region;and(2)how to make wear leveling between SLC region and MLC region.In other words,SLC region should be used to store hot data(frequently accessed or updated data)to benefit from its fast read and write performance and long durability,and MLC region should be used to store cold data to benefit from its low cost and large capacity.Meanwhile,the SLC and MLC region should be synchronous wear as far as possible to realize the same lifespan of heterogeneous SLC/MLC chips.In order to solve these problems,this paper proposes a wear leveling aware flash translation layer(WLAFTL)for hybrid SSDs.Firstly,WLAFTL proposes a dynamic data allocation scheme,which is based on the fusion of wear leveling and write-request size.Specifically,it employs a traditional write-request size based hot or cold data identification and allocation scheme which identifies write requests whose size is less than the threshold value as hot data and sends them to SLC region,and identifies the rest of data as cold data and sends them to MLC region.However,the threshold value of hot or cold data identification is not fixed and can be adaptively adjusted according to the wear speed of SLC and MLC region.Secondly,WLAFTL proposes a garbage collection or migration scheme of cold data for SLC region,which is based on the fusion of wear leveling and first in first out(FIFO)strategy.Specifically,it employs FIFO strategy to manage the data block in SLC region,and when the space of SLC region is not enough,it employs the following scheme to make room for newly arrived hot data:if SLC region wears faster than MLC region,the valid data block of the first in SLC region is immediately migrated into MLC region;otherwise,the valid data block of the first in SLC region is collected into SLC region no more than N times.Our experiment results demonstrate that under the condition of using the same address mapping scheme,compared to ComboFTL and CFTL,WLAFTL can achieve up to 13.6%and 12.7%average response time improvement,as well as 9.2%and 20.4%total block erasure reduction,respectively.Moreover,it can achieve better wear leveling between SLC and MLC than ComboFTL and CFTL.
作者 姚英彪 王发宽 YAO Ying-Biao;WANG Fa-Kuan(School of Communication Engineering,Hangzhou Dianzi University,Hangzhou 310018;School of Telecommunication and Information Engineering,Nanjing University of Posts&Telecommunication,Nanjing 210003)
出处 《计算机学报》 EI CSCD 北大核心 2018年第10期2379-2393,共15页 Chinese Journal of Computers
基金 国家自然科学基金项目(61100044 61671192) 浙江省重点科技创新团队项目(2013TD03) "电子科学与技术"浙江省一流学科A类资助~~
关键词 混合固态硬盘 闪存转换层 磨损均衡 数据分配 冷/热数据识别 hybrid solid-state drive FTL wear-leveling data allocation cold/hot data identification
作者简介 姚英彪,男,1976年生,博士,教授,中国计算机学会(CCF)会员,主要研究领域为存储系统设计、无线传感器网络、媒体信号处理等.E-mail:yaoyb@hdu.edu.cn.联系电话:13282105137.;王发宽,男,1991年生,硕士,工程师,主要研究方向为固态硬盘存储技术.E-mail:w469144949@163.com.
  • 相关文献

参考文献5

二级参考文献105

  • 1顾宝根,顾喜梅.日志结构的嵌入式文件系统研究[J].计算机工程与设计,2004,25(6):915-917. 被引量:17
  • 2Lai S.Flash memories:Successes and challenges[J].IBM Journal of Research and Development,2008,52(4/5):529-535.
  • 3Chang Lipin,Kuo Teiwei.Efficient management for large-scale flash-memory storage systems with resource conservation[J].ACM Trans on Storage,2005,1(4):381-418.
  • 4Park C,Seo J,Bae S,et al.A low-cost memory architecture with NAND XIP for mobile embedded systems[C]//Proc of the 1st IEEE/ACM/IFIP Int Conf on Hardware-Software Codesign and System Synthesis.New York:ACM,2003:138-143.
  • 5Wu M,Willy Z.eNVy:A non-volatile main memory storage system[C]//Proc of the 6th Int Conf on Architectural Support for Programming Languages and Operating Systems.New York:ACM,1994:86-97.
  • 6Taeho K,Trevor M.FlashCache:A NAND flash memory file cache for low power Web servers[C]//Proc of the 2006 Int Conf on Compilers,Architecture and Synthesis for Embedded Systems.New York:ACM,2006:103-112.
  • 7Kgil T,Roberts D,Mudge T.Improving NAND flash based disk caches[C]//Proc of the 35th Int Symp on Computer Architecture.New York:ACM,2008:327-338.
  • 8Dushyanth N,Eno T,Austin D.Migrating server storage to SSDs:Analysis of tradeoffs[C]//Proc of the 4th ACM European Conf on Computer Systems.New York:ACM,2009:145-158.
  • 9Microsoft Corp.Explore the features:Performance[EB/OL].[2008-12-05].http://www.microsoft.com/windows/windows-vista/features/performance.aspx.
  • 10Shmidt D.Trueffs wear-leveling mechanism[R].Newark,CA:M-System,2002.

共引文献105

同被引文献36

引证文献5

二级引证文献11

相关作者

内容加载中请稍等...

相关机构

内容加载中请稍等...

相关主题

内容加载中请稍等...

浏览历史

内容加载中请稍等...
;
使用帮助 返回顶部