Reliability Modeling of Large Fault-Tolerant Systems

A cluster based ultra reliable architecture is pre sented o ering synchronization and system function ality comparable to that of fully connected systems with reduced system overheads Existing combina torial and Markov models do not su ciently model concurrently occurring faults in such large systems A reliability model considering the distribution of con current faults across the system clusters is shown to increase the accuracy of reliability and system fault tolerance estimates The hybrid fault model which classi es faults based on their behavior further im proves reliability estimates and enhances the fault handling capability of each cluster Linear growth in cluster reliability with respect to cluster size is possi ble as are re nements in the convergence and consis tency algorithms for synchronization