An architecture for enabling VNF auto-scaling with flow migration

Network functions virtualization (NFV) together with Software-defined networking (SDN) are complementary technologies that offer elasticity in Network Functions (NFs) control. Due to the changes in network policy, traffic characteristic, or physical topology in Software-Defined NFV (SDNFV) systems, the control plane needs to carry out network updates frequently. Besides, the control plane should enable to orchestrate migrating NF states and redistributing packet processing across multiple instances of NFs to support NFV elasticity control such as NF auto-scaling, NF maintenance. Currently, existing frameworks are proposed to implement these functions separately, thus, there is still a need for a common orchestrator that provides efficient, coordinated control of NF auto-scaling, NF state transfer, and flow migration. In this paper, we discuss the NF states transfer and flow migration method for the high-availability NFV environment. Furthermore, this paper presents a proposal and experimental study of an architecture that supports dynamically state and flow migration to realize NFV elasticity control.