Efficient LRU Algorithm for Cache Scheduling in a Disk Array System

Abstract

Redundant Arrays of Inexpensive Disks (RAID) technology is an efficient way to solve the bottleneck problem between CPU processing ability and I/ O subsystem. Usually, there exist very large amounts of cache memory on the disk array controller board. How to schedule data in the cache so as to enhance the hit-ratio and reduce the time for the disk array to access data from the disks in the array is important for the design of the disk array controller. It is the key factor to measure the performance of the disk array. In this paper, a new cache-scheduling algorithm for the disk array, called track-based improved LRU algorithm, is proposed. The basic element, during the scheduling, is a track of disk drive. It is the combination of read-ahead caching, fast-write, elevator write-back and LRU replacement policy. Track-based improved algorithm uses read-ahead caching as the read policy, fast write as the write policy, elevator write-back as the destage policy and LRU as the replacement policy. Compared to the traditional cache-scheduling algorithm, the performance of disk array can be greatly improved, as the simulation results demonstrate.