| Due to the structural design of mechanical hard disks in the past,their own shortcomings and limitations are gradually exposed in the current development trend of mobile storage technology.On the other hand,solid-state hard drives with NAND Flash chips as the storage core have the characteristics of low power consumption,low heat,anti-vibration,no noise,and small size.They will soon play a very important role in the storage industry.Become one of the storage systems with the most development space and technological prospects.NAND Flash has been used as the main storage chip of the solid state drive,FTL technology is used between the upper system and the underlying storage device to hide the characteristics of the Flash memory,thereby making the solid state drive the I /O Read and write access operations can be as fast as ordinary disk operations.This thesis focuses on optimizing the design of the three modules of address mapping,garbage collection,and wear leveling in the FTL algorithm to improve the performance and storage life of the flash conversion layer.The content of work and the main innovations of this thesis are as follows:(1)In the address mapping optimization algorithm,the length attribute column is added to the cache mapping table,multiple consecutive mapping records are merged into one,and the occupation of the cache slot is reduced.The cache mapping table is removed in batches according to the maximum value of the same global mapping table.The mapping unit reduces the update frequency of the global mapping page and improves the processing capacity for continuous data.(2)Garbage collection generally selects the storage block with the most invalid pages as the recycling block for processing,while wear leveling believes that garbage collection should use the storage block with the least number of wear as the recycling block,so in order to balance this demand,the garbage collection process Design a block value calculation formula that can take into account garbage collection and wear leveling to select the data block that needs to be recycled.This formula takes into account the efficiency of garbage collection and the demand for wear leveling.In this thesis,the first-in first-out algorithm is used as the guiding idea,and the block value calculation formula is used to calculate the storage block that is written first.Then the storage block that is calculated first is likely to be selected as the reclaimed block.The calculation formula takes into account the proportion of invalid pages in the storage block and the number of equal block erases,and this calculation formula can be dynamically adjusted according to the usage and status of the solid state disk.(3)For the static wear leveling mechanism based on thermal data circulation placement proposed in this thesis,this mechanism re-optimizes the current static wear leveling mechanism from reducing the amount of data migration,which can reduce the data migration overhead in the static wear leveling process,reduce the amount of cold data migration and improve the efficiency of cold data migration,so the task of data migration can be completed efficiently and quickly under the condition of a certain amount of data migration,and ultimately can well improve the storage performance of the solid state drive.From the verification results,it can be seen that the continuous I/O performance of the optimized FTL algorithm in this thesis is better than that of the improved DFTL algorithm.The amount of data transmitted per unit time of a full continuous read request is increased by about 17.9%,while the OFTL algorithm is at 4K random I/O.The performance is close to the DFTL algorithm. |
PDF Full Text Request