A Traffic Scheduling Method Based on SDN

Abstract

The rise of big data has brought severe challenges to data storage, which also means that computing resources and storage resources have become centralized. In the past, business data could end up after being processed locally or at the secondary node. Now, all data must be uploaded to each data center for unified processing. Data collection also requires users to access data center services and then send them back, which undoubtedly increases the pressure on the backbone network. This also means that the annual private line cost of the enterprise will increase with the growth of network traffic. Large enterprises often use multiple lines between network nodes to improve link utilization and network reliability. However, in most cases, the line utilization of the backup line is usually low, because the traditional routing algorithm forwards according to the optimal path. Load sharing or traffic scheduling is achieved by adjusting interface cost values ​​or BGP parameters. In large networks, it often affects the entire body. Each adjustment requires detailed planning and validation. At the same time, due to the diversity of applications, different applications have different business characteristics. If the network can distinguish different applications, sense network conditions, and adjust the QoS and traffic paths of applications as needed, the link utilization of leased lines and the network quality of applications will be significantly improved. SDN separates the control plane from the data plane and intelligently controls the network forwarding policy through the global controller, which improves the flexibility and security of the network. Based on the traditional large-scale enterprise WAN architecture, this paper proposes a complete traffic scheduling solution based on SDN to achieve secure, controllable, and flexible scheduling of enterprise WAN network resources.