Improving Reliability, Energy-Efficiency and Security of Storage Systems and Real-Time Systems

Abstract

Many life-critical systems are required to operate without a system failure for a given period of time. Examples are nuclear, aerospace, spacecraft and other such systems. Many of these applications are either storage intensive or real time intensive, hence it is crucial to look at the reliability improvements of computer systems, mainly storage and real time systems. With respect to storage system reliability, in the business arena, data preservation and data mining has proven to be a boon in shaping business strategy. For individuals, storage is being called upon to preserve sentimental and historical artifacts such as photos, movies and personal documents. In both these areas, storage must keep pace with a growing need for efficient, reliable, long term storage. When it comes to real time systems, performance is their most important characteristic--right up until the point where it stops working. Then, suddenly, you don't care how fast it is--or rather, was. You just want it to start working again. The importance of reliability in both storage and real time systems is made clear by the above statements. Reliability of these two computer systems can be affected by a wide range of novel technologies, including energy conservation techniques and security mechanisms. In what follows, we describe two challenging issues of improving reliability in storage systems and real-time systems. The first issue is improving reliability and energy efficiency of storage systems, the second challenge is the improvement of reliability and security in real time systems. Modern day storage systems offer high levels of performance and disk capacity at low costs. In this dissertation, we propose methods for building energy-efficient and reliable large scale storage systems. The primary focus of our research is to achieve the twin goals of maximizing reliability and minimizing energy consumption by incorporating these energy-efficient and reliable techniques to large scale storage systems. The experimental results using both synthetic and real world applications (traces) have shown that the energy consumption could be significantly reduced while guaranteeing maximum reliability for storage disks with a marginal degradation of performance. Reliability of real time systems, second challenging issue, is studied in the later part of the dissertation. For real-time embedded systems we proposed techniques to integrate fault recovery and security services in real time embedded systems. Simulation results show that our techniques can significantly improve security over the conventional approaches, while achieving an efficient means of fault recovery.