Research On Load Balancing Technology Based On GCN And Swarm Intelligence Optimization In Data Center Network

With the widespread application of cloud computing and virtualization technologies,as the infrastructure for various services,the data center network(DCN),as the infrastructure for various services,undertakes increasingly more traffic data.At the same time,users also raise higher requirements for service qualities.To enable data center to provide better services,the key is to design a specific load balancing strategy by combining its network characteristics.To this end,the network load balancing technology for data center is analyzed and researched in this thesis.The major contents are as follows:1.To address the problems of time delay in traffic scheduling and large traffic collision in data center network,a link load balancing strategy based on the graph convolutional neural network(GCN)is proposed.Firstly,a link load state prediction model is designed.Based on the topologic structure and traffic characteristics of data center network,this model uses graph convolutional neural network and recurrent neural network(RNN),to extract the spatial characteristics of data center network and the time-order characteristics of traffic,thereby improved the model’s perception precision of link load.Then,a link load balancing algorithm is proposed,the algorithm measures the link selectivity according to the predicted information of link load state and the actual link information obtained by the controller.With the objective to minimize the collision of large traffic at the core layer,uniform distribution of traffic can be achieved.Next,by combining the comprehensive index of link,the selectivity of path can be determined.An improved artificial bee colony algorithm is introduced to calculate the routing strategy of large flow,and single-dimensional neighborhood search is optimized to multi-dimensional neighborhood search to accelerate the convergence speed.The simulation results show that the proposed load balancing strategy can effectively enhance the load balancing,reduce the transmission delay,and improve the bisection width of the network.2.To address the problems of control expansion and load imbalance under the distributed control plane,a load balancing strategy for cross-domain collaboration is proposed.First,a mathematical model of control resource consumption is established,and specific quantitative analysis of the consumption of control plane resources is carried out.It is can be seen that the inter-domain switch migration method can be used for collaborative management of the control plane resources,the method dynamically adjusts the scope of each control domain according to the number of traffic passing through the control domain,thereby reducing the consumption of control resources.Then,a multi-objective function that minimizes load deviation and minimizes control resource consumption is constructed,to ensure that the method can achieve load balancing on the control plane while reducing control resource consumption.Finally,a cross-domain collaboration algorithm based on NSGA-II is proposed to calculate the solution set of multi-objective functions,the algorithm converts the resource allocation problem into a problem of obtaining the Pareto front,which resolves the conflict between the two objectives.The simulation results show that while reducing the consumption of control resources,this strategy can maintain the load balancing of control plane and improve the network performance.