Improving e-commerce system performance with dynamic system tuning

The World Wide Web has experienced tremendous growth over the past ten years. Along with that growth there have been growing pains that have at times slowed service to a crawl and even disrupted it by crashing web servers. Part of the Web proliferation has been the e-commerce boom. Merchants have been rushing to obtain and maintain a web presence for fear of losing market share to a competitor. Figures from recent studies show the value of consumer e-commerce to be over {dollar}171.4 billion in 1999. Estimates for the future eclipse that figure. Studies show that one of the primary factors causing a consumer to leave a web site in favor of a competitor is excessive delay in receiving information. Much work has been done trying to reduce this latency, but none of these methods allow an e-commerce web system to dynamically alter its configuration to meet quality of service (QoS) parameters without manual intervention or stopping the servers of the site. With this in mind, this dissertation presents a novel method that dynamically adjusts the configuration parameters of an e-commerce site to direct its QoS metrics to desired levels.