Research On Multi-channel I/O Optimization Of Solid-state Storage Arrays

In recent years,storage technology has developed rapidly.In addition to capacity breakthroughs,hardware upgrades have greatly improved storage performance,but its I / O performance has not increased significantly.With multi-core CPU and multi-user mode frequent requests increasing in the multi-channel solid-state storage array of solid-state storage devices,the optimization of its I / O performance is particularly important.This paper designs and implements a multi-channel solid-state storage array scheme using NAND Flash as the storage medium.When there are a large number of frequent I / O requests,the upper layer requests need to pass through multiple different system software layers to reach the lower layer hardware to achieve data access.For this purpose,a dedicated block device driver is designed for the 3-channel solid-state storage array The physical characteristics of the hardware,average access to the commands stored in the SRAM command pool in a polling manner to optimize I / O scheduling,and achieve a more efficient I / O scheduling strategy.Good flash management can better handle I / O requests,and a virtual and real address mapping table is maintained in the Flash Translation Layer(FTL)to implement mechanisms such as address mapping,wear leveling,and garbage collection.The physical characteristics of the internal structure of the flash memory make the use and management of data blocks more intuitive,and make full use of the parallel resources of the flash channel.In addition,the Direct Memory Access(DMA)method is used to transfer data and commands between the driver layer and the solid-state storage array,which can avoid a large number of requests to the CPU and improve the speed and performance of I / O requests.Finally,the analysis and comparison of the system's read and write performance,IOPS,etc.under different sizes of data blocks,different number of channels and different read and write ratios,and show the actual test results.Under different channels,the bandwidth and I / O speed of this system have been greatly improved,in line with the design goals.