Providing Fault Tolerance In Parallel Storage Systems

Report ID: TR-057-86
Author: Balasubramanian, K. / Park, Arvin
Date: 1986-11-00
Pages: 10
Download Formats: |PDF|
Abstract:

Reliability is a critical concern for designers of parallel data storage systems. These systems consist of large numbers of storage devices which can provide high rates of data transfer. However, the consequential dependence on large numbers of devices can make such systems more prone to failure than non-parallel systems which depend on only a single storage device. This problem can be remedied by employing a modest amount of redundant storage to provide fault tolerance. In fact, providing fault tolerance in parallel data storage systems requires less redundancy (and is therefore more cost effective) than providing fault tolerance for non-parallel systems. This paper describes and analyses an effective method of providing fault tolerance in parallel data storage systems.