Rethinking Metropolitan and Wide Area Networks in the Software-Defined Era

Due to the growth of the Internet users and bandwidth dependent applications, the demand for bandwidth will grow over the next decades. Emerging applications such as high quality multimedia and gaming entertainment are driving growth. Moreover, cloud-based services such as virtualization and elastic computing are making Internet traffic more dynamic. Optical networking is a promising technology which is able to address the need for higher capacity. However, to unleash the maximum potential of these networks in the presence of dynamic data traffic, a flexible and efficient network management system is required to optimally utilize the network and avoid unnecessary over-provisioning. Software-defined Networking (SDN) enables the network to be programmable by segregating the control and the data planes in the network equipment. In SDN-enabled networks, control plane is a logically centralized entity that can control and manage all network resources in a more efficient way. It enables network engineers to develop their own network orchestration software that hosts optimization algorithms to efficiently manage the network resources. In this dissertation, the service provisioning and deprovisioning times and the resource utilization are studied to evaluate the performance of SDN-enabled optical networks. Several optimization heuristics and algorithms are proposed that can be implemented in the network orchestration software. These algorithms increase the network throughput in the presence of highly dynamic service requests, and provide optimal cross-layer routing solutions in a survivable network by reducing network over-provisioning.