Huijun Zheng, Wei Xia, Qin Zhou, Yong Peng, Jia Wang, Yanjun Zhu, and Huaide Yang
SDN, load balancing, energy saving, data centre
The rapid development of data centre networks has brought challenges in terms of high energy consumption and quality of service (QoS) assurance. To address the problems of high energy consumption and uneven resource allocation in data centres, a load balancing and energy saving algorithm (LB-ESA) is proposed. First, taking QoS and energy conservation into account, a multi-objective combined optimisation model is established by collecting network bandwidth, delay, device resource utilisation, and temperature through a software-defined networking (SDN) controller. Then, an algorithm is designed to propose an energy-saving mechanism that filters out nodes with excessively high or low utilisation rates, puts idle devices into sleep mode, and introduces an ant colony algorithm for load balancing route search. For the purpose of addressing the problem of ant colony algorithms easily falling into local optima, a logistic chaos model is introduced to disturb the update of pheromones, maintaining population diversity and allowing the algorithm to find a better solution within a limited number of iterations. Finally, an experimental environment is established in Mininet for simulation experiments. The experimental results indicate that LB-ESA outperforms equal-cost multi-path routing (ECMP) and efficient and energy efficient virtual round robin (EEVRR) algorithms by dynamically balancing network load, conserving energy, and maintaining network QoS.
Important Links:
Go Back
