The Simulation of a Highly Dependable Distributed Computing Environment

Abstract

The aim of the research is to investigate techniques that support the development of highly dependable applications in a distributed system environment. Techniques developed include redundant task allocation, load balancing, fault-tolerant computing, and communication. The application the authors have implemented in the system is a firewall application. The firewall runs in redundant mode. Each incoming or outgoing packet is checked by two or more copies of the firewall application. Only when the majority of the firewall copies decide to accept the packet can the packet go through the firewall. Disagreed decision from the different firewall copies signify a possible hardware fault or a software error in the underlying system. This article reports a recent implementation of a simulation system in which computing nodes, redundant copies of tasks, and packet queues are implemented as independent threads, as well as experiment results obtained based on queuing models.