Hierarchical resource model-driven performance analysis with dynamic data control

Abstract

One of the key issues confronting today's telecom carriers is how to rapidly detect variation in service quality to improve customer satisfaction. Since the volume of data required for performance analysis has been increasing due to diversification of network environments, the time required to detect any variation in the network condition has also grown. Due to the complexity of managed data, network services are at a risk of degrading customer experiences. Accordingly, to enable instant performance analysis, we present lightweight performance analysis to dynamically change the monitoring network resource via our proposed two functions. One is a continuous rule-creation function, which creates an analysis rule to change a monitoring resource based on hierarchical network topology and the network condition. The other is a data control function, which can suppress performance data by controlling a database based on the analysis rule. These functions can facilitate not only understanding of network quality, but also reduction of the analysis processing load. To evaluate our approach, a mechanism implemented into a prototype performance analysis system was successfully demonstrated in a network testbed. Its effectiveness was validated in that the proposed performance analysis mechanism reduced CPU utilization by one-third compared to the existing performance analysis approach.