A Dynamic Polling Scheme for the Network Monitoring Problem

Network monitoring is one of the most significant functions in network management. In order to measure network elements and services, normally a polling approach is used to help network professionals understand the state of the managed systems in real-time. Given the current rapid growing networks, the information exchanged between network nodes and a central network management system (NMS) could be substantial. In this study, a dynamic polling scheme is proposed to transform the network monitoring problem into an optimization problem of polling frequencies assignment when network resource consumption becomes a constraint. The motivation for such scheme arises from the needs of network administrators who assign management efforts efficiently on the elements within the networked system. The proposed dynamic polling scheme determines the polling frequencies to each of the network nodes in accordance with the integration of the Analytic Hierarchy Process (AHP) method and Branch-and-bound techniques. Thus, the goal of the optimization is to trade-off between the management efforts on the managed elements and the performance of network behaviours. The simulation results show that the proposed dynamic polling scheme is more efficient than a simple random polling scheme, and performs well across a wide range of traffic loads.