Design and Validation of an Analytical Model to Evaluate Monitoring Frameworks Limits

Abstract

Abstract-It is essential that a monitoring system is being designed with performance and scalability in mind. But due to the diversity and complexity of both the monitoring and the monitored systems, it is currently difficult to reason on both performance and scalability using ad hoc techniques. Thus, both simulation is required and analytical models based on well established techniques such as queueing theory have to be developed. In this paper we provide an analytical modeling of the behaviour of the commonly used manager-agent monitoring frameworks within two scenarios: single manager-single agent and single manager-multiple agents. The two designed models enable the automation of the estimation of the scalability limit of the two types of management monitoring schemes regarding a performance metric like the monitoring delay. We validate our developed models through simulation based on parameters values obtained from the performance measurement of a life running JMX-based monitoring applications with the two scenarios.

On the Selection of Management/Monitoring Nodes in Highly Dynamic Networks

“…The context of the research is management and monitoring. Monitoring frameworks were analysed in [21], [22], [23]. Most proposed monitoring systems are not designed for dynamic networks.…”