Research On Load Balancing Of Fat Tree Data Center Network Based On SDN

With the rapid development of Internet technology in China in recent ten years,a variety of network business applications and large-scale data flows have emerged.At present,under the environment of the rapid rise of cloud computing,the scale of data center network is getting larger and larger,and the internal traffic pressure of the network is also increasing.How to effectively schedule the internal traffic of the network,alleviate the pressure of the flow,and ensure the load balance of the network,has become an urgent problem to be solved in the data center network so far.The use of traditional TCP/IP architectures that are difficult to maintain and scale is no longer sufficient to meet the evolving needs of todayundefineds data center networks,and problems such as low bandwidth utilization persist.Under this background,people research on the new network architecture,so the idea of software-defined network emerges as the times require.The concept of SDN subverts the architecture of a traditional network.in the concept of SDN,the key idea is the decoupling of the network control layer and the forwarding layer,and compared with the distributed control of the traditional network architecture,the software defines the centralized control function on the network logic to greatly increase the network flexibility,And the state information in the network can be acquired more conveniently through the controller in real time.By introducing the SDN technology into the data center network to solve the problem of internal load balancing,this paper focuses on an SDN-based network load balancing strategy of the fat-tree data center,which is widely used as the network structure of the data center.In this paper,Ryu controller is selected as the controller of SDN control level,on which four self-defined modules of load balancing strategy are developed and implemented:topology discovery module,network resource monitoring module,ant colony algorithm routing module,flow table import module,and so on.An ant colony optimization algorithm based on link load balancing is proposed in the routing module of ant colony algorithm,which modifies the original ant colony algorithm.The state of the data center network is regarded as the heuristic factor for the ant to select the path.Therefore,the algorithm can select the current optimal transmission path for the data center network under the condition of multiple constraints.The operation process of the load balancing strategy is as follows: firstly,the topology discovery module acquires the topology information of the whole data center network,and the network resource monitoring module calculates the parameters of bandwidth,delay and packet loss rate in the network.Then the LLBACO algorithm in the routing module of ant colony algorithm uses the information obtained from the previous two modules to select the optimal transmission path,in which the traffic load is the main influencing factor ofant selection path.Delay and packet loss rate are the secondary influencing factors of antundefineds path selection.Finally,the flow table import module encapsulates the current optimal path selected by ant colony algorithm(ACA)routing module into a flow table item and sends it to the OpenFlow switch and completes the corresponding data flow processing decision.The load balancing strategy can dynamically adjust the network load through the real-time network state information obtained by the SDN controller,and realize the dynamic load balancing,so that the network data can be forwarded efficiently.In this paper,we use Mininet network simulation platform and Ryu controller to construct SDN environment,simulate fat tree topology on Mininet,and deploy load balancing strategy in Ryu controller.In this paper,we simulate the real traffic in the data center network of fat tree based on SDN,and carry out three groups of simulation and comparison experiments.The experimental results show that the load balancing strategy in this paper has a good effect on improving bandwidth utilization and reducing transmission delay.