Efficient Decomposition-Based Methods for Optimal VNF Placement and Chaining

The concepts of network function virtualization (NFV) and software-defined networks (SDNs) are emerging as core technologies to satisfy the growing number of user demands in 5G wireless networks and beyond. NFV and SDNs offer network providers the ability to deliver scalable, lower-cost services free from the constraints of hardware components.The virtual network function placement and chaining (VNF-PC) problem is considered in this paper. VNF-PC involves placing virtual network functions (VNFs) optimally in a distributed NFV infrastructure (NFV-I), efficiently allocating network resources to these VNFs to satisfy users demands and directing the network flow through the VNFs. Current solutions to this problem are time-consuming, cannot handle real-time queries, or use heuristic methods to find approximate solutions.To this end, we propose an efficient exact solution method for the VNF-PC problem through decomposition. We show how decomposing the problem can allow us to assess the viability of incumbent solutions and use this information to provide effective cuts. We further augment this exact solution approach by using a greedy branching heuristic to explore more fruitful areas of the search space. Experimental results show the potential of this combined approach.