Optimal fault-tolerant resource allocation in dynamic distributed systems

Abstract

This paper presents a fault-tolerant resource allocation algorithm in a dynamic distributed message passing system, where concurrent processes sharing system resources can be created or terminated dynamically. The degree of fault-tolerance is measured by the failure locality that is the maximum number of processes whose liveness conditions (e.g., starvation freedom) cannot be satisfied because of a single process failure. The algorithm guarantees the optimal failure locality.