Track 2 Session 7
1:00 to 2:00 p.m. Wednesday June 18, 2008

A Method for Quantitatively Predicting the Availability of a Complex Communications Network

Critical communication networks require high availability; however, their complex architectures make availability very difficult to predict. High availability strategies such as the selection of reliable components and redundant architectures have significant cost implications. This session describes a systematic approach to model system availability and direct specific design choices in order to meet a given high availability goal.

The communications network used for this case study is a complex system composed of proprietary and commercial subsystems, leased services and environmental factors, all of which contribute to unavailability. In order to make economic yet quantitatively based design decisions, a method to evaluate the effects of those decisions upon availability was needed. Choosing the "best" components and architecture might not have been sufficient or it might have been costly overkill. With an extensive network, the effect of these decisions was simply too difficult to predict.

This presentation discusses a methodology that strikes a balance between these competing design goals using a logical blend of reliability and availability prediction techniques, reliability modeling software, failure budgeting and conservative estimation. This same approach can be applied to other complex system designs in order to assure cost-effective availability.

Key Words: Critical Communications System, Complex Network, High Availability, Design Guidance