Abstract-- With increasing richness in features such as personalization of content, web applications are becoming more complex and hence compute intensive. Traditional approaches towards the design of web cluster architectures have targeted static content such as images that are usually network intensive. However, these methods are not applicable to dynamic content applications which are more compute intensive than static content. This paper proposes a suite of algorithms which optimize the performance of dynamic content applications by considering both server CPU loads and network latencies. The goal of the algorithms is to reduce client access times while also minimizing the resource utilization. The algorithms are designed for a hosting architecture comprising a grid of clusters inter-connected via high bandwidth links and each cluster hosting a complete application replica. A server migration algorithm allocates servers on-demand within a cluster while a server selection mechanism ...
Supranamaya Ranjan, Edward W. Knightly