Current HTTP servers process requests using a first come first serve queuing policy. What this implies is that the web server must process each request as it arrives. The result is that the more requests a client makes, the more replies the server will generate in response. Unfortunately, the bandwidth of the network and the processing capabilities of the server are often limited resulting in an aggressive client, or sets of clients, consuming the majority of the server’s resources, limiting other clients’ ability to use their fair allocation. While the traditional behavior of a web server works efficiently for a web site that is non-discriminating towards all clients, guaranteeing service for preferred clients from the server itself is not yet possible. This paper describes the algorithm we have designed and implemented on the Apache HTTP server, which has been shown to be effective in allocating configurable fixed percentages of bandwidth across numerous simultaneous clients, ind...