Architecting for Reliability - Detection Mechanisms

Telecommunications networks are evolving towards widely distributed systems of specialized functionality. The primary considerations is continuous availability of service offerings. Future networks will be increasingly comprised of commercial products that lack the specialized detection and recovery that was designed into the previous generations of network equipment. As a result the options for detection and recovery will be limited by the manufacturer and the system's detection and recovery must be designed carefully. To study the impact of detection time on system availability we constructed a six-state Markov model that allows us to vary coverage levels over different detection time intervals. This model also allows examination of the escalating detection, where different detection techniques are used over different timescales. The results confirm some assumptions and also identify the better configurations of high coverage stages in our escalation. Scenarios that exemplify various configurations are discussed to illustrate these results.