A Load Balancing Mechanism for 5G Data Centers
Xin Cui, Qingke Meng, Weihan Wang · 2020
The popularity of 5G technology and related services has caused a sharp increase in network traffic, so there are higher requirements for traffic load balancing within the data center. For 5G data center network resource management and load balancing, a load balancing routing algorithm based on SDN (software defined network) flow scheduling and neural network differential movement is proposed. In this paper, traffic is distributed into Elephant Flow and Mice Flow. At the same time, they are initially routed by different methods, and the movement merit of the variance of link utilization is calculated to determine whether the reroute mechanism requires being activated. Additionally, to set the message cost aside, an improved binary fallback algorithm is used to determine the timing of calculating the variance shift value of link utilization. The routing mechanism includes convection discrimination, radical route and reroute. The routing flow merit obtained by the initial flow and the routing mechanism is used as the input of the training set and the output is used to train the neural network. At the same time, the neural network and the processing of the difference value are used to obtain the equilibrium of the new flow to the entire network.