Research On Software Defined Networking Technology For Data Center

As the core infrastructure of cloud computing,data center networks are bridges that connect tens of thousands,ten thousand or even millions of large-scale server clusters to achieve large-scale distributed computing,so it become a hotspot in the field of Internet research.With the expansion of network scale,data center needs to deal with more information which is also showed an explosive growth trend,some issues become more apparent,such as construction management complexity,control is not flexible and so on,thus affecting the further development of data center networks.However,by using software defined networking technology to build data center not only can simplify the network management,but also can optimize network performance,and further improve data centers business carrying capacity,which has been the industry and academia attention.In this paper,we firstly studied data center networks,software defined networking and the application of SDN technology in data center networks,then focusing on the study of Openflow which is the key technology of SDN,and the forwarding mechanism of messages in Openflow,at the same time,we analyze the existing flow table forwarding strategy based on Ternary Content Addressable Memory(TCAM).During the study we found that Openflow still exist some technical bottlenecks in the actual application,such as the cost and memory problem of TCAM which is widely used in the routers/switches to achieve quickly query.In view of these problems,we proposed using Combined Multi-Dimension Dynamic Bloom Filter(CMDDBF)to replace TCAM to forward and process messages of flow tables.The CMDDBF solves the shortcomings of TCAM,such as lower space utilization and expensive,can effectively realize multi-field data search and processing operations,and adjust the memory space dynamically according to the scale of the flow table.CMDDBF has good flexibility and scalability,therefore it more suitable for query and process operations of flow table.Furthermore,through experimental analysis,CMDDBF has a good performance in reducing false positive compared to other structures.