Research On Intelligent Management Optimization Method Of Read Disturb Perception In Solid State Drives

With the rapid development of the information age,the amount of data is exploding,and big data applications have placed higher demands on the capacity,performance,and reliability of data storage.Solid state drives(SSDs)based on flash memory are widely used due to their high performance and low power consumption.However,flash memory adopts methods such as multi-bit stacking and shrinking manufacturing processes to increase the storage density,which increases the interference of read operations on the data stored in the flash memory,and read disturb gradually becomes the main factor affecting the reliability of the flash memory.Therefore,how to efficiently manage the read disturb to improve the performance and life of the SSD has become a research hotspot.The method of managing read disturb is to migrate the data seriously affected by read disturb to avoid data corruption.The traditional read disturb management algorithm Refresh has two problems that cause a large data migration overhead: first,a fixed read threshold is used to detect the degree of read disturb,so that the data that can withstand more read disturb is forced to migrate,so the frequency of data migration is High;secondly,the data with different read popularity are stored uniformly,so that the degree of read disturb between the data is large,resulting in the migration of data that is less affected by read disturb.Aiming at the problem of high frequency of data migration,a read disturb management algorithm RDD based on real-time interval detection is proposed.Using the type of data that is sensitive to read disturb as an indicator,the detection mechanism of real-time interval reading is used to accurately detect the degree of data affected by read disturb,thereby increasing the number of read disturb and reducing the frequency of migration.Experimental results show that,compared with the Refresh algorithm,the RDD algorithm can reduce the response time of the system by an average of 15.2% and reduce the total number of system erases by an average of 22.8%.Aiming at the problem of large differences in the degree of read disturb between data,a read disturb management algorithm RHDL based on the separation of hot and cold data is proposed.Identify cold and hot data through the read-hotness detection mechanism,and store the cold-read data and hot-read data separately.This can shorten the write intervals experienced by the hot-read data and reduce the interference of the hot-read data on the cold-read data,thereby reducing the difference in the degree of read disturb.Experimental results show that,compared with the Refresh algorithm,the RHDL algorithm can reduce the response time of the system by an average of 18.75% and reduce the total number of system erases by an average of 13.83%.In order to ensure the practicality of the optimization algorithm for read disturb management,a read disturb management algorithm Min RRO that combines the advantages of RDD and RHDL is proposed.Experimental results show that,compared with the Refresh algorithm,the Min RRO algorithm can reduce the response time of the system by an average of 27.75% and reduce the total number of system erases by an average of 30%.