Robust Software Defined Networks Through Dynamic MiddleBox Selection

In the current networks, multiple servers and Middleboxes (MBs) are applied to enhance the network performance and availability. Therefore, different load-balancing strategies are in use to distribute network traffic among these devices. For resource utilization optimization, the status of all network devices should be considered in the traffic distribution procedure. Software-defined Networking (SDN) has an important role in maximizing network resource utilization, availability, and robustness. In this paper, we implemented a "resource-based" load-balancing method in SDN that considers the MBs’ status during the dynamic routing to avoid overloaded MBs in SDNs. Overloaded devices can cause network unavailability and vulnerability. To observe the efficiency of the proposed solutions, we used mininet emulator, and we extracted the CPU, RAM, and bandwidth usage of each MB during the test time. The obtained results show that our proposed method improves the availability and resource utilization of the tested network approximately 85% and 95% in comparison to Round-Robin (RR) and IP-Hash based approaches, respectively. The results also declared that bandwidth and CPU are the most crucial resources of the tested MBs.