Performance evaluation of dynamic supporting algorithms

Abstract

The widely used performance measures for replicated file systems are file availability and reliability. The authors compare such measures for the dynamic supporting algorithm, the dynamic supporting (G) algorithm, and other algorithms, e.g., the available copies and the majority voting algorithms. Since replicas and votes are conceptually separated, the two dynamic supporting algorithms could achieve very high availability and reliability while still keeping storage cost very low, especially with only two replicas. Both stochastic analysis and simulation results are reported. The performance improvement of the dynamic supporting algorithms is demonstrated by the results.<>