混合存储数据迁移策略研究  被引量:2

Research on Hybrid Storage Data Migration Strategy

在线阅读下载全文

作  者:罗保山[1] 张鑫 王栩[2] 谭支鹏[2] 

机构地区:[1]武汉软件工程职业学院计算机学院,湖北武汉430205 [2]华中科技大学计算机学院,湖北武汉430074

出  处:《计算机技术与发展》2016年第6期82-86,共5页Computer Technology and Development

基  金:中央高校财政专项基金(2015XJGH010)

摘  要:大数据环境下海量增长的数据对存储介质的实时响应和超高容量带来了极大的挑战。传统机械硬盘的性能瓶颈日益突出,以固态盘为代表的拥有高性能的新型存储器不断涌现,但其价格远高于机械硬盘。为了降低总体成本同时利用HDD的超大容量,又充分利用SSD的高速性能,以固态盘与机械硬盘构建的混合存储阵列成为一种不错的选择。文中设计实现了一种基于文件级数据迁移的混合存储方案。通过数据分析器统计每个文件的随机访问度和热度,并通过数学模型量化出文件价值来衡量文件的重要性,基于0/1背包模型提出了一种迁移算法,完成文件集合的迁入迁出并实现了对已迁移文件的访问重定向。测试表明,该方案较另外两种存储方案平均提升了10%左右,而且减少了对SSD的写操作,延长了SSD的使用寿命。The explosion of data in the big data environment brings great challenges to the real-time response and high capacity of storage medium. The performance bottleneck of traditional hard disk is becoming increasingly prominent. New type of memory with the high per- formance represented by the solid state disk is emerging constantly ,but its price-performance ratio is not better than that of the hard disk drive. The hybrid storage system that combines HDD with SSD becomes inevitable for reducing the overall costs while leveraging the large capacity of HDD and making full use of high-speed performance of SSD. R analyzes and designs a file-level hybrid storage system in this paper,which uses the data analyzer to count the degree of random access and hotspot for each file. Through establishing the mathe- matical model,it quantifies file value to measure the importance of files,and it uses the 0/l knapsack problem model to design a migra- tion algorithm that helps determine the migration set and realizes the rexdirection for access of each file. Through test, it improves 10% on average compared with the other two storage schemes, reduces the write operations to SSD and prolongs the service life of SSD.

关 键 词:固态盘 混合存储 文件价值 数据迁移 

分 类 号:TP301[自动化与计算机技术—计算机系统结构]

 

参考文献:

正在载入数据...

 

二级参考文献:

正在载入数据...

 

耦合文献:

正在载入数据...

 

引证文献:

正在载入数据...

 

二级引证文献:

正在载入数据...

 

同被引文献:

正在载入数据...

 

相关期刊文献:

正在载入数据...

相关的主题
相关的作者对象
相关的机构对象