Doctor of Philosophy
Electrical and Computer Engineering
Software-defined networking overcomes the limitations of traditional networks by splitting the control plane from the data plane. The logic of the network is moved to a component called the controller that manages devices in the data plane. To implement this architecture, it has become the norm to use the OpenFlow (OF) protocol, which defines several counters maintained by network devices. These counters are the starting point for Traffic Engineering (TE) activities. TE monitors several network parameters, including network bandwidth utilization. A great challenge for TE is to collect and generate statistics about bandwidth utilization for monitoring and traffic analysis activities. This becomes even more challenging if fine-grained monitoring is required. Network management tasks such as network provisioning, capacity planning, load balancing, and anomaly detection can benefit from this fine-grained monitoring. Because the counters are updated for every packet that crosses the switch, they must be retrieved in a streaming fashion. This scenario suggests the use of Big Data streaming techniques to collect and process counter values. Therefore, this paper proposes an approach based on a fine-grained Big Data monitoring method to collect and generate traffic statistics using counter values. This research work can significantly leverage TE. The approach can provide a more detailed view of network resource utilization because it can deliver individual and aggregated statistical analyses of bandwidth consumption. Experimental results show the effectiveness of the proposed method.
Queiroz, Wander, "Big Data for Traffic Engineering in Software-Defined Networks" (2019). Electronic Thesis and Dissertation Repository. 6147.