An Adaptable and Generic Fault-Tolerant System for Distributed Applications

摘要

The work presented in this paper is inscribed within the framework of the autonomic computing which is an initiative started with IBM in 2001. Its final goal is to develop self-managed computing systems and to overcome the rapidly growing complexity problem. In this work, we treated the property self healing of the autonomous systems on account of its importance in the development of the critical systems and with high availability requiring a self-protection against the software failures for ensuring the service continuity. This paper presents GAFTOS (Generic and Adaptable Fault-Tolerant System), a fault-tolerant system adaptable and generic for the distributed and dynamic component-based applications. We illustrate the expressive power of GAFTOS by encapsulating the following fault tolerance techniques in the sub-components of GAFTOS: primary-backup, message store, distributed backward-recovery and distributed check pointing.