Design and implementation of a hybrid prefetching proxy server for WWW

Prefetching systems on the World Wide Web offer considerable improvement of response time; however, a heavy load is put on the host. Aiming at load alleviation, this study considers using an existing prefetching proxy server as the prefetching system. A prefetching proxy server handles client requests while performing prefetching, which implies simultaneous establishment of a vast number of connections. Thus, this study suggests evaluation of load in terms of number of connections, and aims to reduce this number by separation of the two kinds of processing. Unlike the conventional distributed-cache technique using multiple proxy servers, distributed processing is proposed in this study. This makes possible the implementation of a low-load prefetching proxy server while maintaining high-speed operation of WWW services on the client side. In this paper, the load and distributed processing of prefetching proxy servers are discussed, a novel design for a prefetching proxy server is proposed, and experiments and their results are described.