A Lightweight Qdisc Triggered Controller (QTC) for Congestion Aware Software Defined Networks
TL;DR
The experimental findings show that QTC maintains near baseline performance under normal network operation and provides substantial improvements under congestion, indicating that QTC provides a lightweight rule based solution for responsive congestion aware traffic engineering in SDN.
Abstract
Software Defined Networking (SDN) enables centralized and programmable traffic management. However, maintaining service quality under congestion remains a challenging task while using heavy computation or delayed network wide optimization. This paper presents a lightweight Qdisc Triggered Controller (QTC) for congestion aware path switching in SDN. The suggested controller monitors queueing discipline statistics specifically packet drops and qdisc overlimit events. The proposed mechanism is implemented using Ryu controller and evaluated in a Mininet based topology under normal, persistent, transient and sequential bottleneck scenarios. The experimental findings show that QTC maintains near baseline performance under normal network operation and provides substantial improvements under congestion. In the persistent bottleneck scenario, throughput increased from 2.102 Gbit/s with fixed path forwarding to 30.0 Gbit/s with QTC, while ICMP packet loss decreased from 5% to 0%. These results indicate that QTC provides a lightweight rule based solution for responsive congestion aware traffic engineering in SDN.