Resource Orchestration in Softwarized Networks

Abstract

Network softwarization is an emerging research area that is envisioned to revolutionize the way network infrastructure is designed, operated, and managed today. Contemporary telecommunication networks are going through a major transformation, and softwarization is recognized as a crucial enabler of this transformation by both academia and industry. Softwarization promises to overcome the current ossified state of Internet network architecture and evolve towards a more open, agile, flexible, and programmable networking paradigm that will reduce both capital and operational expenditures, cut-down time-to-market of new services, and create new revenue streams. Software-Defined Networking (SDN) and Network Function Virtualization (NFV) are two complementary networking technologies that have established themselves as the cornerstones of network softwarization. SDN decouples the control and data planes to provide enhanced programmability and faster innovation of networking technologies. It facilitates simplified network control, scalability, availability, flexibility, security, cost-reduction, autonomic management, and fine-grained control of network traffic. NFV utilizes virtualization technology to reduce dependency on underlying hardware by moving packet processing activities from proprietary hardware middleboxes to virtualized entities that can run on commodity hardware. Together SDN and NFV simplify network infrastructure by utilizing standardized and commodity hardware for both compute and networking; bringing the benefits of agility, economies of scale, and flexibility of data centers to networks.

Network softwarization provides the tools required to re-architect the current network infrastructure of the Internet. However, the effective application of these tools requires efficient utilization of networking resources in the softwarized environment. Innovative techniques and mechanisms are required for all aspects of network management and control. The overarching goal of this thesis is to address several key resource orchestration challenges in softwarized networks. The resource allocation and orchestration techniques presented in this thesis utilize the functionality provided by softwarization to reduce operational cost, improve resource utilization, ensure scalability, dynamically scale resource pools according to demand, and optimize energy utilization.