Reliability Modeling of Proactive Fault Handling

Abstract

Research on dependable computing is undergoing a shift from traditional fault tolerance towards techniques that handle faults proactively. These techniques comprise two parts: (a) prediction of failures and (b) actions that are performed in case of an upcoming failure. This work provides the first reliability model that incorporates both correct and false predictions as well as both types of actions: failure prevention and recovery preparation. Closed form solutions to availability, reliability and hazard rate are provided.