Placement and Routing of VNFs for Horizontal Scaling

Abstract

To deploy a wide range of services, improve network flexibility and facilitate the deployment and maintenance of network services, virtualization of network functions promises to make the networks more agile and efficient. Network functions virtualization (NFV) enables network service providers to deal with the increase of traffic of data and the deployment of new services by offering a more scalable and flexible network. Many existing studies on NFV have dealt with the placement of virtual network functions (VNFs) in the software layer as a problem of deploying different network function instances within the network without considering routing and horizontal scalability. In the proposed model, we aim to minimize latency, considering horizontal scaling of VMs by placing VNFs across physical nodes. Routing and placement of VNFs are linked and have a significant impact on latency especially for delay-critical services. Therefore, we aim at optimizing the distribution and utilization of available resources and meet user/tenant requirements with the objective of minimizing the overall accumulated latency on both the edge clouds, where VNFs are running and the flow routing paths.