VNF Placement with Optimization Problem Based on Data Throughput for Service Chaining

In service chain technology, each network function is placed as virtual network function (VNF) on the commercial off-the-shelf (COTS) servers. In order to utilize network functions effectively, VNFs can be flexibly added into and removed from the COTS servers. In this case, the data throughput is affected by the amount of CPU resources for each VNF and the bandwidth of each link. In this paper, we propose a VNF placement with an optimization problem to improve the data throughput in service chaining. This method determines (i) the placement of VNFs, (ii) the amount of resources for each VNF, and (iii) the transmission route for each service chain by solving our optimization problem. We solve the optimization problem with CPLEX. We compare our method with Shortest Path (SP) method. From numerical examples, we find that the total data throughput can be maximized by the proposed method. Moreover, it is shown that the proposed method can increase the total data throughput by using the node resources and link bandwidth effectively,