Distributed program reliability analysis

This paper presents an algorithm for computing the distributed program reliability in distributed computing systems (DCS). The algorithm, called FREA (fast reliability evaluation algorithm), is based on the generalized factoring theorem with several incorporated reliability-preserving reductions to speedup the reliability evaluation. The effect of file distributions, program distributions, and various topologies on reliability of the DCS is studied in detail using the proposed algorithm. Compared with existing algorithms on various network topologies, file distributions, and program distributions, the proposed algorithm is much more economic in both time and space. To compute the distributed program reliability, the ARPA network is studied to illustrate the feasibility of the proposed algorithm.<
>