SDN-DVFS: an enhanced QoS-aware load-balancing method in software defined networks

摘要

Recently, software defined networks (SDN) has been considered as a promising technology for improving the network performance. However, the load imbalance problem considerably reduces quality of service (QoS) level in SDNs. Traffic distribution in SDN affects the efficiency and creates other challenges like unbalanced load distribution which will significantly affect the network performance and traffic increase which leads to delay increase as well. To address this challenge, a novel method, named SDN-DVFS, has been proposed to fairly balance the traffic load on the servers and improve the QoS in the network. The proposed method deals with the load-balancing problem in SDNs based on the dynamic voltage frequency scaling (DVFS) in which it considers the overload of each virtual machine (VM), efficiency of the host machine, and the load applied by each user. This method relies on a dynamic traffic in which the on-demand requests arrive one by one without any prior knowledge of future arrivals. SDN-DVFS balances the traffic load over the network and improves the network resource utilization even if there is a large number of VMs in the network. Moreover, the proposed method reduces the synchronization cost between the data and controller layers which leads to the less response time. Regarding energy parameter, the average energy consumption in the proposed method is 1.53 kWh, which is 48.7 less than the number 2.99 recorded by similar method PSOAP. PSOAP considering two parameters of traffic release delay and controllers' capacity as a particle in PSO algorithm adjusts them in SDN in a way that it can improve convergence accuracy and load balancing. Simulation results demonstrated the superiority of the proposed method in terms of the energy, latency, and packet delivery rate in comparison with similar recent methods.