An Approach to Manage Reconfiguration in Fault – Tolerant Distributed Systems

This paper deals with dynamic resource management for real–time dependability–critical distributed systems. Requirements for such kind of systems span many domains such as time, survivability, and scalability and point out formidable challenges in terms of their fulfillment. An architecture is proposed, based on the agent distributed infrastructure Lira, and enriched with statistical models for decision-making capabilities. The aim of the proposed architecture is to provide adaptive system reconfiguration, resorting to a hierarchy of resource managers to cope with fault tolerance and scalability issues.