Reliability of fault-tolerant software based on a system architecture with a recovery metaprogram

The concept of recover metaprogram (RMP) has been proposed as a software unit capable of coordinating the execution of a fault-tolerant software system and performing recovery actions. To evaluate the effectiveness of the RMP in supporting the underlying fault-tolerance mechanism, two well-known fault-tolerant software approaches, N-version programming and recovery blocks, are considered. Markov models are developed to quantify the reliability of the system with or without the presence of the RMP. Preliminary results show that the RMP is capable of improving the reliability of a fault-tolerant software system provided that its intended functionalities are properly used.<
>