Software Defined Network Control Platform Reseach And Implementation

With the rapid development of cloud computing, Internet of things, mobile internet, social networking and other new models, new technologies, new applications, the Internet is also rapidly expanding in scale, the amount of data rapidly increases. The drawbacks of traditional become increasingly apparent, when processing big data, intelligence analysis, network virtualization and other new applications. With bloated complex network structure, rising costs and rigid network management, network researchers have begun to explore new network architecture design.In this environment, software-defined network comes into being. Software Defined Network (SDN) is a new network architecture, which separates the network control from forwarding, providing soft ware programming on network. In SDN, the control platform not o nly simplifies the node devices (nodes are only responsible for forw arding, forwarding behavior controlled by a control platform) but al so network control logic (just call control platform programming int erface, rather than deal with the underlying network status), at the same time supporting a wide range of functions. So the control plat form is a critical component in the SDN architecture. Only an effic ient and scalable control platform can ensure good performance of SDN network, and this is the main research goal of this research.This paper first introduced the origins, architecture and core te chnology of software-defined network, analyzed and compared the e xisting SDN control platform research to define research focus. And then this paper presented a distributed SDN control platform soluti on. The solution implemented the interface between the SDN contro Hers to ensure consistency between multiple controllers’states; only processed important flows to reduce the load on the control platfor m and forwarding devices; initiative issued predefined policies to ea se the control overhead and reduce the initial setup delay of flow.Based on proposed SDN control platform, the paper designed a scalable and efficient data center elephant flow management system. The system only processed elephant flows while using ECMP to handle mice flows, so the system could achieve a good scalability. Meanwhile this paper proposed a weighted multipath routing algorithm, which dynamically calculated the weight for each path and the weights were in accordance with the current network load. Then management system will split detected elephant flows into multiple paths, using the weighted multipath routing algorithm, to balance the load. This paper implemented a system prototype, and verified the effectiveness of the proposed mechanisms in OpenFlow test network. Finally, the paper discussed the future development of SDN control platform.