Study On Allocation Of Virtual Machine Instances And Pricing Mechanism Based On Auction Theory

Cloud computing is a model of providing computing services through the Internet,and its main feature is "using according to the need","extending at any time" and "pay by the usage".Generally,it offer infrastructure as a service in the form of virtual machine(VM)instances,which are characterized by a number of computing resources,such as CPU,memory,storage,and networks.Recently,as technology evolves rapidly and the computing tasks become more complicated and bothersome,more and more users are prefer completing their workloads by the way cloud computing,which will bring huge calculate computation pressure to cloud server and consume more electricity,then lead to more carbon dioxide emissions which has a lasting influence on global climates.It is of great economic importance to study on how to control the consumption of cloud resource effectively.Allocating virtual machine instances effectively and developing price strategy is a research focus in the field of cloud computing.Maximizing social welfare to allocate cloud resources and Pricing with critical value is the two main ideas.Algorithmic Game Theory and Auction Theory provide a new idea for the study of the cloud resource allocation mechanism owns the attributes of "individual rationality" and "incentive compatibility".In recent years,the application of some heuristic allocation algorithm has attracted wide attention.In this context,we study the cloud resource allocation and pricing mechanism to maintain effectiveness,and the main contributions are as the follows:Dynamic Programming based Allocation of Virtual Machine Instances in Clouds..We propose dynamic programming based mechanism for solving Virtual Machine Provision and Allocation problem(DP-VMPA),which takes the maximum social welfare as the objective function,and uses the combinatorial auction-based dynamic programming(CA-DP)allocation algorithm to solve the Winner Determination Problem(WDP).Finally,the Vickrey-Clarke-Groves(VCG)mechanism is used to decide the payment of each user.We prove that the proposed scheme has good economic properties:individual rationality,incentive compatibility and computational complexity.We perform simulation experiments to compare our proposed mechanism with CA-PROVISION mechanism and show that the DP-VMPA mechanism can allocate VM resources more effectively,and bring higher profits to the auctioneer.Research on Virtual Machine Allocation Based on XOR bidding language.We consider the complex user set,that is,the user could submit multiple bid requests to obtain resources to in each auction,but the auctioneer for each user up to only one bid.We introduce the XOR bid language,improve the allocation of cloud resources based on the local search algorithm,and use the improved VCG mechanism to calculate the price of virtual machine resources.This mechanism allows users to fully express the request intention,prove high user experience.Simulation experiments show that this auction mechanism can bring higher revenue for the auctioneer,while giving users a higher production efficiency.Allocation Based on Time Consideration in Online Virtual Machine Auction Mechanism.We have improved the online auction mechanism based on the allocation of virtual machines,taking into account the changes in the demand of the peak and the trough period in the allocation scheme of the mechanism,so that the auctioneer can dynamically measure the bid density of the user's bid at different times,decide the winners more reasonably,improve resource utilization,and bring higher profits to the auctioneer.We tried to use the real workload file to perform lots of simulation experiments to verify the effectiveness of the improved online auction mechanism.