An architecture for real-time active content distribution

The phenomenal growth of the world-wide web has made it the most popular Internet application today. Web caching and content distribution services have been recognized as valuable techniques to mitigate the explosion of web traffic. An increasing fraction of web traffic today is dynamically generated and therefore intrinsically difficult for replication using present static approaches. Scalable delivery of such active content poses a myriad of challenges, including content replication, update propagation, and consistency management. This paper makes two contributions: (1) it describes a scalable architecture for transparent demand-driven distribution of active content; (2) the system can provide real-time delay guarantee on content access. Our approach involves migrating the scripts which generate the dynamic web traffic, and their data, from the server to active content distribution proxies nearest to the clients and mechanisms on these proxies to enforce real-time delay guarantees for client requests. Our system is implemented and deployed on PlanetLab [19], a real-world distributed Internet testbed. Experimental data show that significant improvements are observed in effective throughput and client response time, and delay bounds on content access can be guaranteed with a very high probability.