| With the continuous expansion of the data center scale,the data center traffic increases sharply and the demand for network traffic takes diverse forms.Traditional network management methods are unable to schedule data center network traffic flexibly and efficiently,resulting in lower overall performance of data center network.Software defined network technology has the characteristics of mastering the global view,separation of control and forwarding,and centralized control.It provides a new idea for solving the problem of data center network traffic scheduling.And the application of this technology to the data center can effectively solve the traffic scheduling problem of data center network.Based on software-defined network technology,this thesis has launched an in-depth study on the data center network traffic scheduling problem.The main research contents are as follows:1.To address the problem of unbalanced load and wasted link bandwidth resources in data center networks,this thesis proposes a traffic scheduling strategy based on particle swarm optimization algorithm,which calculates the scheduling path for network traffic through the process of finding the optimal position of particles.Firstly,the strategy optimizes the basic particle swarm algorithm.And then in order to ensure the reasonable use of bandwidth resources,the strategy designs the fitness function by considering the two factors of link real-time status and bandwidth adaptation rate.Finally,the strategy calculates the optimal scheduling path for network traffic by the optimized particle swarm algorithm.The experimental results show that the proposed strategy effectively improves the network throughput and link utilization,and obtains lower packet loss rate.Overall,the strategy improves the overall performance of the network to some extent.2.Data center network contains a variety of traffic with different business requirement,but the existing scheduling strategy cannot be well adapted to it,and most of the strategies are static in the way of traffic classification.To address this problem,this thesis proposes a traffic scheduling strategy oriented to business requirement.Firstly,the network traffic to be scheduled is dynamically divided into three types of traffic with different business requirement based on the available bandwidth of the link and the actual bandwidth demand of the network traffic.Then,considering the link bandwidth,delay and congestion risk and other related factors,the strategy designs the path calculation method corresponding to the traffic demand,so as to complete the effective scheduling of the network traffic.The simulation results show that the proposed strategy in this thesis achieves a higher average network throughput,and effectively reduces the packet loss rate and average round-trip delay. |
