Abstract:
The reliability of multiprocessor computing systems is analyzed taking into account the diagnostic and recovery process characteristic. General construction principles of a model representing the reliability behavior multiprocessor systems and algorithms for implementing this model in statistical simulation are developed. The model takes into account the effect of completeness of testing of the model components and the depth of fault location, of the reconfiguration procedure parameters, and of the reserve capacity. An example is given of the design of a reliability model of a specific system and the results are presented as recommendations that indicate the relative significance of individual characteristics.