基于随机游走的大容量固态硬盘磨损均衡算法

基于闪存的大容量固态硬盘(SSD)能够在未来取代磁盘.它有很多优点,包括非易失性、低能耗、抗震性强等.然而,基于NAND闪存的存储块自身存在有限的擦除重写次数的问题一直影响着它的广泛应用.当闪存芯片达到擦除重写的限制次数后,存储块上的数据就会变得不可靠.目前研究者们已经提出了一些磨损均衡算法来解决这个问题.但当固态硬盘的存储容量不断增大后,这些算法需要越来越多的内存容量来保证运行.文中提出一种基于随机游走的磨损均衡算法来应用在大容量的固态硬盘上,该算法能够很大程度地减少内存消耗.实验表明所需内存容量仅为BET算法的15.6%,与此同时磨损均衡的性能并没有降低.

A Random-Walk Based Wear-Leveling Algorithm for Large-Capacity SSDs

Large-capacity flash based SSD(Solid-state disk) has the potential to become the storage system alternative in the future.It has many advantages: non-volatility,low-power consumption and shock resistance and so on.However,reliability is still a critical problem when using NAND flash memory.Every block of NAND flash memory has a limited number of write/erase cycles,after the limited number,the data that keep in the block become unreliable.Many wear-leveling algorithms have been proposed to solve the reliability problem.But with the capacity of SSD increases,these algorithms always need more and more DRAM capacity.We propose a random-walk based wear-leveling algorithm for large-capacity SSD that can obviously reduce memory consumption(only 15.6% DRAM space compared to BET algorithm) for storing wearing information and achieve the same performance compare with other algorithms.