Research On Performance Optimizations Of Solid-state Disk Storage Systems

To date,solid state disks(SSDs)have been widely used,but how to further apply the advantages of SSDs and to overcome their own shortcomingsis still of great importantance.In this paper,the aim is to improve the life of SSDand to enhance the access performance of external memory by using the dominant performance of SSD;The optimal parallel channel number of SSD and the relationship between the channel number and parallel performance have been determined through simulation drive.Moreover,the dominant performance of SSD has been used for optimizating the structuresof alloy clusters.This paper focuses on four aspects of SSD as follows.1.A new wear-leveling algorithm based on dynamic threshold is proposed.The algorithm adopts the dynamic threshold wear-leveling algorithm,and makes the last period of wear-leveling threshold gradually decreasesby continuously reducing the wear-leveling threshold in the cycle;When the wear-leveling enters the last cycle,the each piece in the SSD has just been worn out;Moreover,each block is almost worn out to death at the same timewithout any differences,which could make the erasable timesof each blockin the SSD be fully utilized.This improves the life of SSDand achieve the goal of performance stability.The experimental results show that the proposed algorithm may remarkably improves the lifetime and stability of the SSD.2.A data placement method based on statistical hybrid storage is proposed.This method USES the statistical method to obtain the access characteristics of the data and divides the data into hot data and cold data.Putting hot data into SSDS and cold data into HDDS takes advantage of both media.At the same time,the cold and hot features of the data that has been put into SSD were extracted,and the data that has been cooled in SSD were written back to HDD.The simulation results show that the hybrid storage system can obviously improve the response time of data access and the life of SSD,and also reduce the cost of storage.3.Analyze and optimize multichannel SSD access.By simulating the performance of the solid-state disk with different channel Numbers,the performance data generated after simulation is compared,and then the analysis finds the best number of parallel channels that can give full play to the performance of the solid-state disk.It is also found that the number of parallel channels is not linearly related to the performance of the solid disk.4.In this scheme,the solid-state disk is applied to the optimization calculation of alloy cluster structure algorithm to further improve the computational acceleration ratio.Using Spark to optimize the structure of CoPt cluster under the parallel framework requires some data dumps in the memory.In the process of dump,data of different features should be reasonably placed in different external storage.The approach in this article is to place dump data that will still be brought into memory for use in the future directly on SSDS with better performance,and dump data that will not be used in the future directly on SSDS.By comparing the calculation using only HDD data dump placement method,the data dump placement method after adding SSD storage is more conducive to calculation,and makes the acceleration ratio significantly improved.