An Improved Framework of Disaster-Tolerance Oriented Adaptive Failure Detection

Abstract

The detection of failure is a fundamental component for highly reliable systems. Taking the QoS of failure detection into consideration, a Framework of Disaster-Tolerance Oriented Adaptive Failure Detection (DTO-FDF) is presented, which adopts hierarchical modularity design and constructs the DTO-FDF and algorithm from the system view and data circulation sequence. DTO-FDF mainly includes three modules: Monitoring module, Processing module, Responding module. By monitoring and gathering the data of node, according to evaluating index and configuring policy, combining with the designed failure detection algorithm, the system judges the machine is alive or not. Based on the result of decision, the system makes the take-over response and the system migration. The framework made some benefic work for the study of failure detection module of universality and independence.