Dynamic scheduling of multiclass open queueing networks in a slowly changing environment

The popularity and importance of Web have increased dramatically in the past few years as well as the complexity of Web server systems. Workload characterization studies reveal that there exist strong time-of-day effects in the Web traffic. Many Web sites have sustained and higher hit rates during certain time periods of a day than other time periods. During the peak hours, the Web servers may even be overloaded. Simple stochastic processes with a fixed rate fails to capture this time varying characteristic of the Web systems. Therefore, we herein consider that the Web system is operating in a changing environment. Whenever the environment changes state, the arrival rates of user requests change as well as the service rates and the routing decisions.