Fault-Tolerant System Availability and Rewards Model with Validation
01 January 2019
The recovery and repair durations of large fault-tolerant systems gen-erally span several orders of magnitude. The distributions also violate the com-mon modeling assumption of an exponential distribution for the recovery and repair time. A reward based semi-Markov model is presented that can be used to predict the steady-state availability of such systems as well as evaluate de-sign trade-offs with respect to their impact on system availability. The model has been validated against field outage data from a large system.