Energy-efficient virtual network function placement based on metaheuristic approaches

• Autor(a) principal: Mosaiyebzadeh, Fatemeh
• Data de Publicação: 2020
• Tipo de documento: Dissertação
• Idioma: eng
• Título da fonte: Biblioteca Digital de Teses e Dissertações da USP
• Texto Completo: https://www.teses.usp.br/teses/disponiveis/45/45134/tde-13082020-200722/
• topic: Cloud computing; Computação em nuvem; Eficiência energética; Encadeamento de funções de serviço; Energy efficiency; Network functions virtualization; Service function chaining; Virtualização de funções de rede

Concerns about reducing energy consumption in the sector of Information and Communication Technology has increasingly motivated the transition of traditional services to the clouds. In this context, Network Functions Virtualization (NFV) emerges as a solution to migrate various network functions, from dedicated hardware devices to a virtual environment based on commodity hardware. With this virtualization, in addition to the promise of increasing energy efficiency, it is expected to reduce the financial cost and increase the flexibility and scalability of the networks. In this research, it is proposed the development of algorithms based on three metaheuristics (Standard Hill-Climbing, Simulated Annealing, and Memetic Algorithm) to schedule network functions in cloud data centers, observing not only the capacities and energy consumption of the computers where the functions will be executed but also of the network and switches that connect these computers. Comparing the algorithms proposed in relation to the Best Fit algorithm found in the literature, the one based on Simulated Annealing saved 55.44% of energy consumption in a datacenter with Three-tier topology and the one based on memetic algorithm saved 49.18% of energy consumption in a data center with Fat-Tree topology. To allow the reproduction of all the experiments carried out in this research, the codes developed are publicly available as free software