Trade-offs encountered in the design of fault-tolerant systems

Early work on fault-tolerant computing was aimed solely at increasing the dependability of computer systems given the possibility of hardware failure. As complexity increased, it became evident that various type of systematic failure were possible, and that these must also be accounted for by the designer of a dependable system. This paper explores the question of whether fault-tolerance techniques can be successfully employed to deal with both random and systematic failures and shows that several trade-offs are encountered