CLRU: a new page replacement algorithm for NAND flash-based consumer electronics

Abstract

NAND flash memory has become the dominant secondary storage device for consumer electronics. However, existing data structures and algorithms in modern operating systems are optimized for magnetic disk-based storage device in light of its mechanical nature and NAND flash memory shows different hardware characteristics from magnetic disk. Therefore, a new page replacement algorithm should be designed for NAND flash-based storage device. In this paper, a new page replacement algorithm is proposed for consumer electronics equipped with NAND flash memory as secondary storage device. The proposed algorithm reduces the number of write operations to NAND flash memory by delaying the eviction of cold dirty pages and improves the page hit ratio by evicting the cold pages preferentially. Experimental results show that the proposed algorithm is better than existing page replacement algorithms designed for NAND flash memory in terms of page hit ratio and the number of write operations.