一种固态硬盘混合损耗均衡算法研究

发布时间：2018-05-07 02:11

本文选题：闪存 + 固态硬盘　；参考：《国防科学技术大学》2012年硕士论文

【摘要】：由于闪存独特物理特性的限制，闪存存储块的擦除次数不能超过一个上限，否则存储在里面的数据就会不可靠。因此需要采用损耗均衡算法将固态硬盘的磨损均衡到整个存储空间中，从而延长固态硬盘的使用寿命。但是损耗均衡算法会引入额外的擦写操作，并且随着大容量闪存最大可擦除次数的降低，损耗均衡算法所引入的额外擦除次数带来的影响逐渐突显出来。本文为解决大容量固态硬盘控制器中损耗均衡算法引入额外擦除次数较多的问题展开研究，主要做了以下几方面的工作：（1）介绍了闪存及固态硬盘的特点和损耗均衡算法在闪存存储系统中的重要作用，分析了目前损耗均衡算法的研究现状，指出了损耗均衡研究的热点和未来发展趋势，并通过分析对比选择了混合损耗均衡算法进行改进。（2）针对混合损耗均衡算法额外擦除次数较多的问题，提出了一种区分组内组间策略的改进方法，将资源消耗小的随机性损耗均衡策略应用于分组内部，将均衡效果好的确定性损耗均衡策略应用于分组之间。在组内，通过控制组内随机性损耗均衡的触发频率和提高冷数据选择的精度降低了混合损耗均衡算法组内损耗均衡操作带来的额外擦除次数。组间采用阈值控制损耗均衡策略，提高了组间损耗均衡操作的针对性，，降低了混合损耗均衡算法组间损耗均衡操作带来的额外擦除操作。提出的改进算法利用固态硬盘并行总线这种特殊结构，进一步降低了混合损耗均衡算法所需要的内存空间。（3）设计了一系列的仿真和测试实验，对提出的改进混合损耗均衡算法的均衡效果、额外擦除比例、内存消耗和处理器资源消耗等性能进行了仿真测试，仿真和测试结果表明提出的改进算法能够保持损耗均衡的效果和所需的内存开销基本不变，同时将额外擦除比例从1.7降低到1.2。
[Abstract]:Due to the limitations of the unique physical characteristics of flash memory, the number of erasures of flash memory blocks cannot exceed one upper limit, otherwise the data stored in it will be unreliable. Therefore, it is necessary to use loss equalization algorithm to equalize the wear of solid state hard disk to the whole storage space, so as to prolong the service life of solid state hard disk. However, the loss equalization algorithm will introduce additional erasure operations, and with the reduction of the maximum erasable number of large capacity flash memory, the influence of the additional erasure times introduced by the loss equalization algorithm is gradually becoming more and more obvious. In order to solve the problem of introducing more extra erasure times into the loss equalization algorithm of solid state hard disk controller with large capacity, this paper mainly does the following work: This paper introduces the characteristics of flash memory and solid state hard disk and the important role of loss equalization algorithm in flash memory system, analyzes the present research status of loss equalization algorithm, and points out the research hotspot and future development trend of loss equalization. The hybrid loss equalization algorithm is improved by analyzing and comparing. 2) aiming at the problem of more erasure times in hybrid loss equalization algorithm, an improved method to distinguish the inter-group strategy is proposed, which applies the random loss equalization strategy with low resource consumption to the intra-group. The deterministic loss equalization strategy with good equalization effect is applied between groups. In the group, by controlling the trigger frequency of random loss equalization in the group and improving the precision of cold data selection, the extra erasing times of the hybrid loss equalization algorithm are reduced. The threshold control loss equalization strategy is adopted among the groups, which improves the pertinence of the inter-group loss equalization operation and reduces the additional erasure operation caused by the hybrid loss equalization algorithm. The improved algorithm further reduces the memory space required by the hybrid loss equalization algorithm by utilizing the special structure of the parallel bus of the solid-state hard disk. A series of simulation and test experiments are designed to test the performance of the improved hybrid loss equalization algorithm, such as equalization effect, extra erasure ratio, memory consumption and processor resource consumption. The simulation and test results show that the proposed algorithm can keep the loss equalization effect and memory overhead unchanged, and reduce the extra erasure ratio from 1.7 to 1.2.
【学位授予单位】：国防科学技术大学
【学位级别】：硕士
【学位授予年份】：2012
【分类号】：TP333

【参考文献】