—Processors speed have increased in a steady pace year over year, storage system performance is still the major bottleneck for most computer systems including high performance computers. Many attentions have been given to optimize the storage system speed using various approaches such as caching, intelligent perfecting and scheduling techniques, nevertheless, the storage system remains the performance bottleneck for most computer systems. Solid State Devices (SSD) have lately been used as a cache layer located between the system main memory and the magnetic hard drives in order to create robust and cost effective hybrid storage systems. The reason comes from the growing density of the SSDs at lower prices with main advantage of high random read efficiency compared to magnetic hard drives. These new devices are capable of producing not only exceptional bandwidth, but also random I/O performance that is orders of magnitude better than that of standard rotating mechanical disks due to the absence of moving parts. In this paper, we have conducted an extensive empirical and comparative study of an I/O intensive workload running on hybrid storage system. We have configured an SSD-Aware real system with variable RAM, SSD, Working Sets configurations in order to evaluate the performance gain achieved by utilizing the SSD device as a middle layer between the RAM and the Hard Disk. This attractive middle layer has also motivated us to propose and simulate new SSD-Aware Hybrid Caching Architecture (HCA) that utilizes an SSD as an extended read cache to the main memory. We have developed a Hybrid Cache Simulator to explore the design space of the hybrid cache using both performance and cost metrics and test it for two I/O intensive real system workloads. Our simulated architecture along with the real system experiments have shown that SSD can effectively be used as a cache extension to the main memory to minimize the disk hits ratio that would otherwise cause substantial delay in workload performance.
Index Terms—Hybrid storage system, multi-level caching, data intensive application, solid state devices, SSD-aware page replacement policy.
A. Aldahlawi is a PhD candidate at Electrical and Computer Engineering Department, The George Washington University Washington DC, 20037, USA (e-mail: email@example.com).
E. Al-Araby and T. El-Ghazawi, and S. Suboh are with the Department of Electrical Engineering at The Catholic University of America, Washington, DC 20034 USA (e-mail:firstname.lastname@example.org, email@example.com, firstname.lastname@example.org).
Cite: A. Aldahlawi, E. El-Araby, S. Suboh, and T. El-Ghazawi, "An Empirical and Architectural Study of Using an SSD-Aware Hybrid Storage System to Improve the Performance of the Data Intensive Applications," International Journal of Information and Electronics Engineering vol. 2, no. 5, pp. 720-730, 2012.