A Practical Model for Optimal Placement of Virtual Network Functions

Abstract

Network Function Virtualization (NFV) can reduce significantly capital and operating costs for communication service providers by shifting network functions from dedicated hardware to Virtual Network Functions (VNFs). VNF placement problem is one of the challenges in NFV. Current studies are limited in the practical applicability. To tackle this problem, we study the optimal VNF placement in NFV with symmetrical, two-way traffic and reusable VNF instances. The problem is formulated as a Mixed-Integer Linear Programming (MILP) model. Then, a simulation is conducted using a real-world network service to show the performance of the proposed model.