The Research On Cloud Computing Profit Model Based On Multiserver System Configuration Strategy

In a cloud computing structure,there are three layers:infrastructure providers,service providers,and customers.Infrastructure providers supply basic physical resources and software facilities.Service providers rent resources from infrastructure providers and provide services to customers.Customers submit their service requests to service providers and pay for them according to the workload and service quality.We pay our attention to profit from cloud service providers' perspective,since they play a critical role in cloud computing.The profit is impacted greatly by the configuration of a cloud service platform,which is usually modeled as a multiserver system with infinite capacity.Adopting this configuration scheme,service providers might provide services of low quality for customers due to great demands and limited computing resources.According to the service level agreement between customers and service providers,low-quality services will decrease the profit to certain extent.To improve the quality of service and increase the profit,we propose a novel multiserver system configuration scheme in which a finite-capacity scheme and a double-renting strategy are combined to provide improved services to customers and to increase revenue and profit.The innovations and contributions are as follows:We use two kinds of rental schemes to reduce the resource waste,which are fixed-rental scheme and temporary-rental scheme.Combination of the two renting schemes leads to higher utilization of resources.A finite-storage queuing model is adopted in our paper.In this model,the capacity of system queue is limited.When the number of the queuing service requests reaches the maximal capacity,temporary servers are rented to execute the newly arrived service requests.This strategy gives a simple but effective way to improve the quality of service when service requests have their own deadlines.In this paper,each task has a deadline which is proportional to the number of instructions and the maximal system capacity is limited.Each entering request is firstly placed in a finite-capacity waiting queue,and if the queue is full,temporary servers are rented to execute the newly arrived service request.Based on the profit equation,a complete searching method is adopted to obtain the optimal solution,i.e.,the optimal configuration of server size,speed,and capacity,such that the profit is maximized.The proposed scheme is compared with the infinite-capacity scheme from several aspects.Comparison results show that our scheme outperforms the infinite-capacity scheme in terms of service quality and profit.