Abstract: With the advances of Peer-to-Peer (P2P) technology, streaming service was widely used over the Internet.However, the dynamics of overlay topology as well as the spontaneous behavior of users make it challengeable to improve the service quality.To address the above challenges, we design and implement a web monitoring system based on NJUVoD, an interactive P2P streaming service platform developed by Nanjing University.By using collaborative filtering mechanism and clustering algorithm, the web monitoring system could analyze the user behavior, topology structure and system status on-the-fly.In addition, it provides a tool for understanding the runtime status and optimizing the system performance.