Metrics, methodologies, and tools for analyzing network fault recovery performance in real-time distributed systems

Abstract

The highly distributed computing plants planned for deployment aboard modern naval ships will serve real-time, missioncritical applications. The high-availability requirements of the distributed applications targeted for these platforms require mechanisms to rapidly recover from system faults (e.g., battle damage). Providing network fault recovery mechanisms that support rapid recovery in the network infrastructure used to compose these systems is critical. This paper examines how metrics used for general network performance analysis can be used to analyze fault recovery performance. A methodology for applying these metrics is presented. A testing toolset that implements the metrics and complies with the testing methodology is presented. Finally, test data collected using the toolset is presented to show the utility of the metrics and testing methodology for evaluating network fault recovery performance. © 2000 Springer-Verlag Berlin Heidelberg.