Research On Perturbation-based Load Balancing Routing And Traffic Scheduling Algorithm Under SDN

With the rapid development of cloud computing,big data and other Internet applications,data center becomes a huge information exchange center since it stores and manages massive business and data.How to effectively improve the bandwidth resource utilization is the key issue facing the data center network.Due to the highly dynamic of data center traffic,it is difficult to maintain load balancing only rely on the route calculated by the network initial state.Therefore,applying adaptive traffic scheduling strategy according to real-time load status of network is an important way for data center network to realize dynamic load balancing.This thesis mainly studies the problem of routing and traffic scheduling when network is perturbed(the network link utilization variance exceeds the threshold).The main contents of this thesis are as follows:(1)This thesis proposes a dynamic load balancing routing algorithm based on perturbation.Due to the shortcomings of selecting the same link and the deficiency of selecting route only according to the sum of link utilization,this thesis uses the comprehensive utility value which is calculated by the sum and the variance of link utilization.Considering the network perturbation,this thesis divides the full links into light and heavy link queues,to achieve the goal of equilibrium.Based on the OpenFlow environment built by Mininet and Floodlight,this thesis carries out the simulation experiments of D-LBR.Simulation results show that D-LBR can allocate data traffic more balanced than other algorithms,and improve the network link utilization and throughput in the initial routing process.(2)We propose a difference weight PSO based on crossover genetic.Considering the fact that the traffic scheduling is sensitive to delay when the network is perturbed and the shortcomings of the exhaustive method in traditional routing algorithm,this thesis maximizes the utilization of optimal value and improves the convergence speed and seeking accuracy of the algorithm through the guidance method.(3)This thesis presents an adaptive traffic scheduling algorithm based on perturbation.Taking the network perturbation into account,we first filter the traffic object of rerouting based on the threshold.Then put forward the fitness function based on variance criterion and comprehensive utility value,take D-LBR as the initial route calculation method and combine DWGPSO for combinatorial optimization to find the optimal solution of the rerouting paths,implement the fast rerouting process and realize load balancing.Simulation results show that ATS has better performance criteria than other algorithms under the condition of network perturbation.