Research and design of a self-managing storage cluster

Abstract

As a result of the Internet's rapid growth, there has been a huge increase in the amount of information generated and shared by people all over the world. Demand for large-scale storage system is growing rapidly. Instead of disk arrays, decentralized storage systems that consist of a cluster of nodes with locally attached disks have become a cost effective solution. This paper describes the design and implementation of SMSC (self-managing storage cluster) - a self-managing high-performance PC-based cluster storage system. SMSC virtualizes distributed storage devices as incrementally expandable volumes and automatically manages storage node additions and failures. The cache algorithm aims at multimedia data, which greatly improves the performance of sequential read, which is especially suitable for multimedia applications. This paper presents experimental results to demonstrate features and performance of SMSC using benchmarks.