Research On Data Center Network Congestion Control Algorithm Based On SDN

At present,cloud computing and big data services are growing vigorously,and large data centers are also developing rapidly.The data center traffic has exploded,and the internal network topology is becoming more and more complex,and the scale is getting bigger and bigger.Traditional congestion management has gradually exposed some deficiencies.The emergence of software-defined networking(SDN)provides new solutions to phenomena such as link congestion,load imbalance,and link redundancy.The difference between SDN and traditional networks is reflected in the decoupling of the forwarding plane and the control plane,which is conducive to network congestion management.Compared with the traditional network,the control layer under the SDN architecture can use programming methods to control the behavior of the data layer,which is considered to be an effective method to manage the complex network environment.Nowadays,managing network congestion based on the SDN architecture,managing the complex and huge traffic in the data center,and making full use of network link resources have become a hot issue in network research.The specific idea of this thesis is to use the centralized control method of SDN to give full play to the multipath characteristics of the data center network topology,and to further optimize and solve the problem of link congestion by improving the rerouting flow scheduling congestion control algorithm.The specific work of the thesis is as follows:(1)The topology measurement is realized through the link discovery protocol under the SDN architecture,and the connection relationship between the switches is provided for the congestion scheduling algorithm design.The switch port counter function realizes bandwidth measurement,and provides network link bandwidth information for subsequent algorithm design.(2)A priority-based flow scheduling congestion management algorithm(CC-PK)is designed.When the link is congested,the scheduling flow is first selected from the congested link based on the scheduling priority,and then in the reachable short path The rerouting scheduling path is selected based on the path priority,and the large flow rerouting on the congested link is transferred to other idle paths to alleviate the congestion.(3)A flow scheduling congestion management algorithm based on simulated annealing is designed.When congestion occurs,by setting the scheduling flow set and backup path set,the simulated annealing algorithm is used to obtain a link load that can reduce the link load below the congestion threshold.According to the flow scheduling scheme,the flow is rerouted and scheduled according to the scheme to relieve congestion.Finally,the simulation results show that the two algorithms can effectively alleviate congestion while improving link utilization and data stream transmission performance.