TaBLA: a client-based scheduling algorithm for Web proxy clusters

As client populations in ISPs continue to rise, it becomes necessary far ISP proxy caches to efficiently handle large numbers of Web requests. We examine the performance of client-side load balancing schemes that help select a proxy from an array of proxies that are equidistant from the client. The current most popular solutions include choosing a random proxy based on either the URL requested, or the Web server from which the URL is requested. Based on an analysis of proxy traces, we propose a new client-side scheduling algorithm "TaBLA". The algorithm creates a redirection table that can be loaded into the clients' browsers. Trace-driven simulations indicate that our algorithm significantly improves average response time and average slowdown compared to the purely randomized schemes.