Performance-reliability improvement in a distributed system

Abstract

An assessment of an optical-fiber LAN (local area network) in which some strategies for increasing fault tolerance have been defined is presented. By means of a stochastic model based on a Markov chain with an absorbent state, the basic parameter linked to network reliability is assessed. In particular, the increase in mean time to failure and performability achieved by introducing a new monitor recovery process is determined. The results presented confirm the validity of the monitor recovery process in enhancing the fault tolerance of a token-ring network and show that the strategy for defining fault-tolerant protocols and architectures effectively enhances network reliability.<>