This paper presents DONet, a data-driven overlay network for peer-to-peer live media streaming. DONet operates by periodically exchanging data availability information between nodes and retrieving or supplying data from partners. It emphasizes three key features: ease of implementation, efficiency in data forwarding based on availability, and robustness through adaptive switching among multiple suppliers. Analysis shows that DONet is scalable with bounded delay. Practical challenges are addressed with efficient membership and partnership management, and an intelligent scheduling algorithm for real-time streaming. DONet was evaluated on PlanetLab, demonstrating good streaming quality under challenging network conditions with low control overhead and transmission delay. A public Internet-based implementation, CoolStreaming v.0.9, was released, attracting over 30,000 users with more than 4,000 online simultaneously. It successfully broadcasts live sports programs, reducing server load and improving user experience. The paper discusses related work, including tree-based and gossip-based protocols. DONet's design avoids complex structures, using data availability to guide data flow. It features a membership manager, partnership manager, and scheduler. The system handles dynamic node behavior, with a scalable membership and partnership management algorithm and an intelligent scheduling algorithm for efficient streaming. Performance evaluations in PlanetLab show that DONet maintains high streaming quality and continuity, even under dynamic conditions. It outperforms tree-based overlays in terms of playback continuity and has comparable delay. CoolStreaming, a practical implementation, demonstrates the effectiveness of DONet in real-world scenarios, with results showing that larger overlays improve streaming quality. The paper concludes that DONet is a promising solution for live media streaming, with potential for further improvements and optimizations. It highlights the importance of overlay-based streaming in overcoming limitations of traditional client-server systems and the need for further research in this area.This paper presents DONet, a data-driven overlay network for peer-to-peer live media streaming. DONet operates by periodically exchanging data availability information between nodes and retrieving or supplying data from partners. It emphasizes three key features: ease of implementation, efficiency in data forwarding based on availability, and robustness through adaptive switching among multiple suppliers. Analysis shows that DONet is scalable with bounded delay. Practical challenges are addressed with efficient membership and partnership management, and an intelligent scheduling algorithm for real-time streaming. DONet was evaluated on PlanetLab, demonstrating good streaming quality under challenging network conditions with low control overhead and transmission delay. A public Internet-based implementation, CoolStreaming v.0.9, was released, attracting over 30,000 users with more than 4,000 online simultaneously. It successfully broadcasts live sports programs, reducing server load and improving user experience. The paper discusses related work, including tree-based and gossip-based protocols. DONet's design avoids complex structures, using data availability to guide data flow. It features a membership manager, partnership manager, and scheduler. The system handles dynamic node behavior, with a scalable membership and partnership management algorithm and an intelligent scheduling algorithm for efficient streaming. Performance evaluations in PlanetLab show that DONet maintains high streaming quality and continuity, even under dynamic conditions. It outperforms tree-based overlays in terms of playback continuity and has comparable delay. CoolStreaming, a practical implementation, demonstrates the effectiveness of DONet in real-world scenarios, with results showing that larger overlays improve streaming quality. The paper concludes that DONet is a promising solution for live media streaming, with potential for further improvements and optimizations. It highlights the importance of overlay-based streaming in overcoming limitations of traditional client-server systems and the need for further research in this area.