A Design and Implementation of Network Billing System on Campus Based on Hadoop and Netflow<bold> </bold>

There are many problems in traditional network billing systems basing on traffic, usage time or monthly payment. For example, it is infeasible to calculate huge traffic of 10 Gbps or difficult to achieve multiple connections for single user in diverse network environments. To address this problem, the paper designed and implemented a new Network Billing System (NetBS) on campus based on Hadoop and Netflow. NetBS captures the mirrored traffic from the backbone port of the campus network and converts the traffic into Netflow v5 streams. These streams are stored into Hadoop big data platform in real-time. In this way, the paper conducted precise 1: 1 sample for multi-type core switches and routers and implemented zero-error differential pricing for single user with multi-connections under 10 Gbps traffic. In addition, the paper designed Billing Query and Management Subsystem (BQMS) to manage users' traffic and bills efficiently. The paper deployed NetBS in the campus network of Xi'an Jiaotong University and demonstrated that NetBS is applicable for multi-network scenarios such as ten Gigabit dual-stack link and flexible network traffic billing. NetBS is currently running in the campus network and providing wired and wireless campus network billings for more than 40,000 students in dormitories and teaching areas.<bold> </bold>