Improving Hybrid Flash-Disk Storage through Intelligent Prefetching: A Data Mining Approach

Abstract

In this paper, we describe a sequential pattern-based prefetching method that can be used for hybrid flash-disk storage device, a combination of hard disk and flash memory. Before we employ the hybrid storage as the primary persistent storage device, we have to develop high-level management skills fit for the device. The hybrid storage can use the flash memory as a cache space to improve system performance. With the memory architecture, we focus on prefetching file objects to be accessed in the near future through the sequential pattern data mining technique. Through intensive experiments using UCC and synthetic data, we show that the proposed method achieves to improve system I/O speed by more than 23%.