Research On Key Technology Of Topology Management For Software Defined Networking

With the advancement of science and technology,network traffic has increased dramatically over the past decade,which has made it difficult for traditional network architectures to meet today's needs.As a new type of network architecture,Software Defined Networking(SDN)has the characteristics of high controllability,easy maintenance,and fast recovery,which has attracted widespread attention.The complete decoupling of the network data plane and the control plane are the main feature of SDN,which can well solve the problem that the existing traffic in the traditional network is difficult to control,and the new functions are difficult to extend.As a result,SDN has received high attention from relevant experts and companies.The SDN controller stores a full-network topology,which enables the controller to allocate network resources from a global perspective,thereby improving network utilization and better managing the network.Therefore,how to use the characteristics of SDN to efficiently manage the network is a very worthwhile research.This paper relies on the National High-Technology Research and Development Program("863" Program)project "Software-Defined Network Architecture and Key Technologies" to conduct research on distributed controller deployment,network topology discovery,and Web-side topology display.From the level of the entire network,the level of the sub-network,and the level of the display,the problems at different levels of the network are improved one by one,and finally the purpose of better managing the network topology in the SDN is achieved.The research work and major contributions of this paper are as follows:1.For the unreasonable deployment of controllers in distributed SDN,this paper proposes an adaptive allocation strategy for multi-domain and multi-controllers.The new method integrates the original controller deployment plan into two phases: sub-domain planning and optimization of the number of controllers in the sub-domain.First of all,according to the key indicators such as the degree of attraction and the degree of attribution,clustering,operations are performed on network nodes to achieve reasonable network partitioning and controller location selection.Then,the corresponding cost is nonlinearly interactively integrated to comprehensive delay,failure degree,and communication overhead.The total cost of the sub-network and the number of controller deployments are optimized.Through the simulation test,the proposed method can effectively reduce the transmission delay compared with other methods,and the overall performance is improved by 13%.2.Aiming at the situation that the controller is overloaded during topology discovery,this paper proposes a topology discovery optimization method based on a hash algorithm.After the controller uses the traditional topology discovery method to obtain the network topology and partition the subnet topology,using this method can effectively reduce the load of the controller during subsequent maintenance of the network topology.These method inserts new rules into the switch to make it copy the Link Layer Discovery Protocol(LLDP)sent by the controller.When the LLDP is forwarded through the switch port,it carries a unique Hash MAC tag.The mapping table matches the Hash MAC to find out the switch number and port number that sent out LLDP and discover the link between the two switches.The simulation test shows that the topology discovery method proposed in this paper is 11% lower than the traditional topology discovery method.3.The display topology of the controller graphical user interface is unreasonable,unfriendly,and the presentation methods of different controllers are not uniform.This article designed an operator interface based on the controller northbound REST API.The interface uses a mature front-end framework named Sencha ExtJS to build the operation interface,and combines the jTopo toolkit to draw the topology.In addition,the addition and deletion of terminal nodes are also realized,and it can issue warning when traffic overload.After testing,the interface can accurately and reasonably display the network topology in the actual network,and all functions are performed correctly.