Research On Optimization Strategy Of Flashcache Based On Data De-duplication

Due to the SSD has the advantages of the high access speed and low energy consumption, it has been used in more and more situations, and the hybrid storage system using SSD+HDD to build a hierarchy storage system is developing very fast. However, SSD is limited by the number of writes, and it will result in the problem of reliability. According to the duplicate data in the disk, adding the data de-duplication function to the hybrid storage system will reduces the number of writes in SSD, than the lifetime of SSD will be longer.According to the deeply studying of the SSD's advantages and disadvantages, the structure of open source project flashcache and the related knowledge of data de-duplicate, design a hybrid system based on the data de-duplicate. The overall goal of the system is based on data de-duplicate, before the data is written to the SSD, test the duplicate data is exist in the SSD or not and the duplicate data will not be written to the SSD, then to reduce the number of data block to SSD. According to the main target of the system, the optimization scheme of the flashcache is designed in detail. Design the new I/O scheduling process and collecting dirty blocks mechanisms according to the redesign of the metadata structure. In order to realize the system functions, the new system add the fingerprint database, virtual address information and virtual_fact address pair table to the metadata structure, use the new index structure to solve the I/O requests. By modifying the flashcache sourece to achieve a new scheduling process, complete the data de-duplication in the system and ultimately reduce the number of writes to the SSD, improve the read and write hit rate, thus increasing the SSD's service life and indirectly increase the storage space. Finally, modify the source complete the operation of the system implementation.Based on the framework of flashcache, the improved flashcache is tested about the access speed and hit rate. Through test the performance between the native flashcache system and the new system, we can find the new system can improve hit rate of hybrid storage system from 5% to 10%, reduces the number of writes SSD and prolong the service life of the SSD.