Energy-efficient virtual network function placement based on metaheuristic approaches
Fatemeh Mosaiyebzadeh · 2020
Concerns about reducing energy consumption in the sector of information and communications technology have increasingly motivated the transition of traditional services of this area to the clouds.In this sense, Network Functions Virtualization (NFV) comes as a solution to migrate various network functions, from dedicated hardware devices, to a virtual environment.With this virtualization, besides the promise of increased energy efficiency, it is expected a reduction of the financial cost and an increase of the network flexibility and scalability.However, bad placement decisions of the Virtual Network Functions(VNFs) on the physical hosts can avoid the expected reductions in energy consumption.In this research, we proposed the development of three VNF placement algorithms based on metaheuristic approaches to place the network functions in physical machines of cloud data centers.The first aim of our approaches is declining the energy consumption of all the physical devices involved, including switches and other interconnection elements.Moreover, in this dissertation, we showed the energy consumption of Three-Tier and Fat-Tree data center by applying our three VNF placement algorithm.On average, in simulation experiments, comparing our proposed algorithms with the best-fit algorithm in a simulation environment, the one based on simulated annealing saved 55.44% of energy consumption in a three-tier data center and the one based on a memetic algorithm saved 49.18% of energy consumption in a fat-tree data center.Furthermore, we showed that for service function chaining, the Three-Tier data center is more energy-efficient than the Fat-Tree data center.The second contribution of this dissertation is to develop an open-source framework.