Research On Traffic Optimization In SDN-Based Data Center Networks

In recent years, cloud computing is developing rapidly. And the users of Internet online service continue to increase. As a result, the traffic in data centers of large enterprise grows a lot, and the competition of network bandwidth resources is fierce. Existing forwarding mechanism in data centers is randomly selection from multi equivalent pathes based on static hash, which does not take into account the load conditions of links and may lead to congestions in the network. More reasonable strategy should do flow scheduling based on the current state of the network. SDN is innovative network architecture, separating the control plane from the data plane. Control functions are concentrated in specialized controllers, and forwarding devices are only responsible for data forwarding according to the rules get from controller. In SDN environment, the controller can maintain the network view, and understand the real-time status of the network, making it possible to achieve better data center traffic scheduling strategy.This paper studies traffic optimization in SDN-based data center networks. The main work and research results include:1) Proposed SASCD, network architecture of data center with multi controllers. Then analysised the key steps of traffic optimization in SDN-based data center networks.2) For data centers with a single controller, this paper proposed a genetic algorithm for static flow scheduling. The algorithm was compared with existing hash-based routing and a greedy algorithm. The results have shown that the genetic algorithm can make the load of links more balanced.3) In data center networks with multi controllers, this paper presented a distributed algorithm for dynamic flow scheduling. The results have shown that the distributed algorithm is better than hash-based routing, but slightly worse than the genetic one. However, the distributed algorithm has higher time efficiency, and it's more suitable for large-scale data center networks.The work and research results can provide a basis for further development of SDN-based data center networks.