Evaluation of Fault Tolerance Latency from Real-Time Application's Perspectives

The Fault-Tolerance Latency (FTL) deened as the time required by all sequential steps taken to recover from an error is important to the design and evaluation of fault-tolerant computers used in safety-critical real-time control systems. To meet timing constraints or avoid dynamic failure, the latency of any fault-handling policy | that consists of several stages like error detection, fault location and recovery | must not be larger than the Application Required Latency (ARL), which depends upon the controlled process under consideration and its operating environment. We evaluate the FTL while considering various fault-tolerance mechanisms and use the evaluated FTL to check if a fault-handling policy can meet the timing constraint, FTL ARL, for a given real-time application. The FTL is dependent on the underlying fault-handling mechanisms as well as fault behaviors during the application of temporal-redundancy recovery such as instruction retry or program rollback. We investigate all possible fault-handling scenarios and represent FTL with several random and deterministic variables that model the fault behaviors and/or the capability and performance of fault-handling mechanisms. We also present a simple example to demonstrate the application of the evaluated FTL in real-time systems, where an appropriate fault-handling policy is selected to meet the timing requirement with the minimum degree of spatial redundancy. Any opinions, ndings, and conclusions or recommendations expressed in this paper are those of the authors and do not necessarily reeect the view of the funding agencies.