Performance evaluation of cloud e-marketplaces using non preemptive queuing model

With the drift of consumers to the Cloud e-marketplaces looking for affordable and cost effective services, waiting time is of interest to every consumer and also a key source of competitive advantage for any cloud e-market provider. Keeping consumers´ waiting may incur high costs of consumer´s dissatisfaction, such as loss of future business and actual processing costs of complaints. The evaluation of performance impact on consumers´ waiting time has not been fully addressed in the context of differentiated service provisioning. In this research, we conducted a modelling and evaluation of a typical cloud e-marketplace using two classes of consumers under non preemptive priority discipline. We studied and compared the performance impact of these two classes in regard to the non-priority discipline. Our approach used the analytical and simulation model. Early results revealed that the average total waiting time is independent of the service discipline. Furthermore, unlike the Non-Priority where the service distribution of the two classes are almost equal when the server utilization increases, the class 1 consumers´ waiting time approaches finite limit and that of class 2 consumers deviated slightly from the average total waiting time under the Non preemptive Priority.