Mirror site organization using a social insect metaphor

The increasing obligation of the Internet now means that, it is a common practice to move server content closer to groups of clients through the use of proxies and mirrors. This naturally decreases latency and server load. To do so, some researchers have been working on automatically clustering network clients on a changeable network topology without human interference. In this study, an approach making use of a social insect metaphor and the potential function clustering algorithm is employed to identify automatically the connectivity pattern between clients and servers without predefined constraints. In order to ease the selection of mirror sites for application servers on computer networks, this approach distributes search activities over ants working in parallel. This feature greatly speeds up search process and increases the accuracy of results. The mirror site located at “center” of a cluster is shown to decrease the traffic load and latency of Japanese backbone under different load configuration.