一种固态盘的读写性能优化调度方法

Read-Write Performance Optimization Scheduling Scheme for SSD

相比于传统机械硬盘,基于NAND Flash的固态盘由于具有非易失性、高性能、低功耗等优点,被广泛应用于数据中心、云计算、在线事务交易等场景。然而,由于NAND Flash中的读操作速度远远快于写操作速度,当读写请求并发执行时,读请求可能被写请求阻塞,从而表现出极大的读延时。在许多以读请求为主的场合,尤其是在线事物交易中(读请求占总请求的比例超过90%),读延时的急剧增加严重影响了系统的整体性能。提出一种读写性能优化调度的策略,通过在闪存转换层之下动态调整读写请求的优先序列,使读性能获得显著的提升。实验中,通过对固态盘仿真器的设计与实现,对读写调度策略的有效性进行了系统的评估。实验结果表明,在该调度策略下,系统中读延时的最大值和平均值均得到了显著的减少,且降幅分别达到了72%和41%。

Compared with traditional hard disk drives （HDDs）, NAND-Flash-based solid-state drives （SSDs） are nonvolatile and can provide better performance as well as lower power consumption. Therefore, they have achieved extensive application in data centers, cloud computing and online transaction trading, etc. However, in NAND Flash memory, the speed of read operation is significantly faster than the write operation. Hence, for a concurrent workload with a mixture of read and write requests, reads may be blocked by writes, which exhibites an enormous read latency. In many read-intensive applications,especially the online transaction trading, in which the proportion of read requests is than 90 %, the sharp increase of the read latency influences the overall performance of the system severely. In this paper, we proposed a read-write performance optimization scheduling scheme which achieves remarkable improvement about the read performance by dynamically adjusting the priority sequence of read and write requests beneath the flash translation layer. In the experiment, we designed and built an SSD simulator to evaluate the effectiveness of the scheduling scheme. Experimental results show that by implementing the proposed scheme, the maximum and the average read latency in the system are substantially reduced,with the reduction of 72 % and 41%, respectively.