Reliability estimation of fault-tolerant systems: tools and techniques

A comparative evaluation of state-of-the-art tools and techniques for estimating the reliability of fault-tolerant computing systems is presented. The theory of reliability estimation is briefly reviewed. Five current approaches are compared in detail: HARP (hybrid automated reliability predictor), SURE (semi-Markov unreliability range estimator), HEIRESS (hierarchical estimation of interval reliability by skewed sampling), SHARPE (symbolic hierarchical automated reliability and performance evaluator), and SAVE (system availability estimator). Particular attention is given to design limitations imposed by underlying model assumptions, on the one hand, and the efficiency and accuracy of the solution techniques employed, on the other hand.<<ETX>>