Time-based CFLRU: A Novel Cache Replacement Algorithm for Flash Memory

摘要

Recently, flash memory makes a tendency to replace magnetic disk for the secondary storage of mobile devices. Therefore, it is necessary to re-investigate various operating system polices such as cache replacement policy to optimize them for flash memory-based storage. Operating systems with flash memory need to consider the number of write operations instead of only cache hit rate. In this paper, we propose a new cache replacement policy, called Time-based Clean First Least Recently Used (TB CFLRU) using 5-minute rule. TB CFLRU applies the idea of 5-minute rule where pages referenced every five minutes should be memory resident. In performance evaluation, we compare TB CFLRU with two well-known cache replacement algorithms: Least Recently Used (LRU) and Clean First Least Recently Used (CFLRU). Experiment result shows that the proposed cache replacement algorithm has lower page miss cost and page write cost comparing to LRU and CFLRU.