Analysis of Performability for Stochastic Models of Fault-Tolerant Systems

Abstract

Performability, a composite measure for the performance and reliability, may be interpreted as the probability density function of the aggregate reward obtained from a system during its mission time. For large mission times we show that known limit theorems lead to an asymptotic normal distribution for the aggregate reward. For finite mission times and Markovian models we obtain the expressions for all moments of performability and give recursions to compute coefficients involved in the expressions. We illustrate the use of the results through an example of a multiple processor computer system.