An Effective Solution Based on Genetic Algorithm for Virtual Network Functions Placement

Abstract

Network functions virtualization (NFV) is a new design paradigm for network architecture. It is a potential technology to solve the current issues that the traditional networks face, such as excessive expenditure, complex configuration and waste of resources. NFV can decouple the network functions from physical dedicated hardware and virtualized them as software. Therefore, these virtualized functions, called as virtual network functions (VNF), can be managed flexibly in a cost-effective way. One of the most interesting topics is how to place VNFs in physical networks efficiently. In this work, we investigated this VNF placement problem and explained it in detail. Then we proposed an effective solution based on the genetic algorithm (GA) to cope with it. In order to obtain the expected economic benefits when processing service function chains (SFC), the objective of our solution is minimizing the number of VNF instances implemented. This objective can also help prevent over-provisioning or under-provisioning and achieve reasonable physical resource allocation. The simulation results show that the proposed solution has a good performance in resources utilization. It can process SFC requests effectively by instantiating not too many VNF instances. The delay of the end-to-end service delivery is also controllable when searching for the optimal solution.