An Analysis of Geographical Push-Caching

Most caching schemes in wide-area, distributed systems are client-initiated. Decisions of when and where to cache information are made without the beneet of the server's global knowledge of the usage patterns. In this paper, we present a new caching strategy: geographical push-caching. Using the server's global knowledge and a derived network topology, we distribute data to cooperating servers. The World Wide Web is an example of a wide-area system that will beneet from distance-sensitive caching, and we present an architecture that allows a Web server to autonomously replicate HTML pages. We use a trace-driven simulation to evaluate several competing caching strategies. Our results show that geographical push-caching reduces bandwidth consumption and sever load by the same amount as web proxy caching, but with a savings in global cache space of almost two orders of magnitude. More importantly, servers that wish to reduce Internet bandwidth consumption and their load can do so without waiting for web proxies to be implemented worldwide. Furthermore, geographical push-caching helps distribute server load for all web servers, not just the most popular as is the case with proxy caching.