Improvement And Optimization Of Load Balancing Based On OpenFlow

With the rapid development of big data, cloud computing and network virtualization technology, the traffic in the network has increased dramatically. In order to avoid network congestion and improve the fault tolerance and robustness of network systems, the research of load balancing technology becomes crucial. To some extent,Load balancing technology can improve network nodes the situation of uneven distribution of traffic, some nodes overload, while the other nodes in the idle state. Due to the complex network configuration of the traditional network architecture, the load balancing flexibility is poor, and the global network resources can not be used for traffic scheduling. Therefore, it is urgent to propose a new network architecture for innovation.OpenFlow protocol based software defined network, the core idea is separation of the control plane and the forwarding plane. SDN architecture breaks the hierarchy concept of traditional network, and realizes the global network control in a flat way.OpenFlow controller is located in the control plane, mainly to achieve centralized management of the network, develop forwarding strategy. It provides the API for application calls upwards and downwards through the standard southbound interface protocol. OpenFlow switches are located in the forwarding plane, and are mainly used for high-speed data forwarding based on the forwarding decision mechanism of the controller. The decoupling of the control plane and the forwarding plane provides the flexibility to control the global network and improve the effectiveness of load balancing.To this end, this paper presents an improved load balancing scheme based on OpenFlow.In this paper,for the typical Fat-Tree network topology in the data center, two key issues affecting load balancing are identified: the location of service requests and the search for the shortest path. Then the two problems are improved and optimized, and then a new load balancing scheme is proposed. Firstly, according to the global network topology information and the server load, a wildcard rule AWR is designed to allocate the service request and obtain the matching information between the requesting host and the processing server. The core idea of the rule is to request aggregation according to the server weight priority, and to minimize the flow rules. Then, an improved algorithm AD based on the original Dijkstra algorithm is proposed according to the obtained pairing information, and the shortest path can be found by routing multiple paths.The key of the algorithm improvement is to make the routing strategy by considering the information such as the bandwidth and weight of the bottleneck, and then optimize the bandwidth utilization and delay.At the end of this paper, we build a Fat-Tree topology on the virtual experiment platform Mininet, and test the different scenes to get the experimental data. Then according to the bandwidth utilization and packet round-trip delay performance indicators, verify the superiority of the proposed load balancing scheme.