In this paper, we consider software system optimization design with Recovery Blocks considering uncertainty in component reliability estimate. The design objective is to maximize an estimate of system reliability and also minimize the variance of the reliability estimate. Recovery Block is one of the most common fault-tolerant system architectures, and it is applied for system redundancy, if needed, in this system optimization research effort. We present an optimization model where the system has a serial configuration, and each of the subsystems has choices of with or without RB/1/1 redundancy; a single software fault and a single hardware fault are tolerated. The model is designed under cost constraints. Our model can be easily applied for other types of fault-tolerant system architecture. This is the first time that a technique to optimize reliability of a system using multiple software versions with different reliabilities and correlated failures is proposed. We believe that our reliability optimization of redundant systems consists of realistic assumptions of failure correlation between/among software versions.
展开▼
