Adaptive Aggregation of Flow Records

keywords: Network traffic monitoring, IPFIX, exporter, flow record, data reduction, adaptive aggregation, heavy-hitter, resource utilization
This paper explores the problem of processing the immense volume of measurement data arising during network traffic monitoring. Due to the ever-increasing demands of current networks, observing accurate information about every single flow is virtually infeasible. In many cases the existing methods for the reduction of flow records are still not sufficient enough. Since the accurate knowledge of flows termed as ``heavy-hitters'' suffices to fulfill most of the monitoring purposes, we decided to aggregate the flow records pertaining to non-heavy-hitters. However, due to the ever-changing nature of traffic, their identification is a challenge. To overcome this challenge, our proposed approach -- the adaptive aggregation of flow records -- automatically adjusts its operation to the actual traffic load and to the monitoring requirements. Preliminary experiments in existing network topologies showed that adaptive aggregation efficiently reduces the number of flow records, while a significant proportion of traffic details is preserved.
mathematics subject classification 2000: 68M10, 68M12, 90B18, 90B20
reference: Vol. 37, 2018, No. 1, pp. 142–164