Notice of Violation of IEEE Publication Principles: Dynamic Embedding and Scheduling of Service Function Chains for Future SDN/NFV-Enabled Networks

Abstract

Currently, software-defined networking (SDN) and network function virtualization (NFV) are the two most promising approaches for implementing network virtualization (NV). Traditional TCP/IP-based networks (e.g. Internet) will be transformed into SDN/NFV-enabled networks in the foreseeable future. Through virtualization, the slicing of heterogeneous underlying resources will provide an agile and customized virtual network (VN) services to end users. Hence, the virtual network function (VNF) embedding and scheduling are crucial to the VN service deployment in the SDN/NFV-enabled networks. In this paper, the dynamic VNF embedding and scheduling are jointly researched for enhancing the VN service provisioning. At first, the VNF embedding and scheduling are formulated by using the mixed integer linear programming (MILP) model, having the goals of minimizing the consumed underlying resources and providing QoS-guaranteed VN service. Subsequently, to remove the NP-hardness of the MILP model, a dynamic VNF embedding and scheduling algorithm is proposed. For instance, when a new VN service is requested, the VNFs, constituting the service function chain (SFC) of the VN service, will be embedded and scheduled by the proposed algorithm. If the resource and QoS requirements of the VNFs are not satisfied, a re-embedding and re-scheduling scheme will be triggered in order to optimize certain existing VNFs. The dynamic embedding and scheduling algorithm has flexible network function placement and improves the underlying resource utilization. Finally, the simulation results are illustrated to validate the proposed algorithm.