Closing the Generation Gap: Satisfying Servers, Providers, and Visitors of Hyper-Content

We consider the problem of building a robust and efficient infrastructure for hyper-content. Existing server-only architectures are susceptible to flash crowds, while other peer-to-peer (p2p) solutions may not yet be economically viable because of their voluntary model. We start with the assumption that any infrastructure has to satisfy three types of key players - the content provider, the server, and the visitor. Our solution allows the provider to specify a pricing function, optimizes resource usage and avoids flash crowds at the server, and keeps response times low for the visitor. The solution adapts at run-time in order to satisfy these requirements. With low visitor concurrency, our solution reduces to a server-only solution. With rising concurrency, more and more hyper-content gets distributed and shared out among visitors and providers. This leads us to believe that the "right" infrastructure for hyper-content may lie somewhere in between the current two extremes of server-only and p2p