Abstract

During the past decades, NAND flash memory based SSDs have been widely deployed in personal computers, mobile devices, and cloud systems. In order to maintain the mappings between logical addresses and physical addresses, a built-in RAM called as mapping cache is always implemented within SSDs. However, due to the cost and space limitations of RAM, only limited number of address mappings are supposed to be stored in the mapping cache. The mapping cache has been designed to exploit the access characteristics of traditional file systems, such as EXT4. Recently, flash friendly file system (F2FS) has been widely used and optimized for SSDs, which exhibits completely different access pattern and has not been taken into consideration during the design of mapping cache. In this paper, we are the first in proposing an efficient F2FS aware design for optimizing the efficiency of mapping cache. The basic idea of the work is to exploit the specific access characteristics of F2FS and propose to improve the hit ratio of the mapping cache. Experimental results show that this approach is able to significantly improve the performance of SSDs.

Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call