Bounding transient and steady-state dependability measures through algorithmic stochastic comparison

Open Webpage

Ana Busic, Jean-Michel Fourneau

Published: 2006, Last Modified: 22 May 2024SIGMETRICS Perform. Evaluation Rev. 2006EveryoneRevisionsBibTeXCC BY-SA 4.0
Abstract: We are interested in bounding dependability measures like point and steady-state availability and reliability of systems modelled by very large Markov chains which are not numerically tractable. We suppose that the state space is divided into two classes, UP (system is operational) and DOWN states. The reliability at time t is defined as the probability that the system has always been operational between 0 and t. The point availability is the probability that the system is operational at time t, and the steady-state availability is the limit, if it exists, of this probability.