Improving availability bounds using the failure distance concept

Abstract

Continuous-time Markov chains are commonly used for dependability modeling of repairable fault-tolerant computer systems. Realistic models of non-trivial fault-tolerant systems easily have very large state spaces. An attractive approach which has been proposed to deal with the largeness problem is the use of pruning-based methods which provide error bounds. Using results from Courtois and Semal, a method for bounding the steady-state availability has been recently developed by Muntz, de Souza e Silva, and Goyal. This paper presents a new method based on a different approach which exploits the concept of failure distance to better bound the behavior out of the non-generated state space. The proposed method yields tighter bounds. Numerical analysis shows that the improvement is typically significant.