Research On Cooperative Game In Cloud Federation Platform

The emergence of cloud computing opens the third IT revolution, with the mature ofcloud computing technology, more and more users begin to use the cloud services, cloudcomputing has profoundly changed the way people access to the software and hardtwareresources. In order to promote the healthy and rapid development of cloud computing,enterprises located in the cloud computing industry chain need to cooperate, and provideservices through the way of building ecosystem. Cloud providers play an important rolein the cloud computing industry chain. Multiple cloud providers have complementaryadvantages, so they can provide high-quality service to users through cooperation. Whencloud providers form a coalition, they face a key problem how to distribute the revenue.Based on this, this paper study the cooperative behavior of Multiple cloud providers withthe method of cooperative game, and present revenue allocation for the cloud federation.First, we set up a cooperative model for multiple cloud providers whose internaldemand is fixed, and show that the cooperative game is super-additive and possess anon-empty core. We study a suitable revenue allocation by means of the core, theShapley value, the Owen value and the consensus value. Then we analyze the instabilityof internal demand, and build a cooperative model for multiple cloud providers whoseinternal demand are stochastic. We show that the cooperative game possesssuperadditivity and non-emptiness of the core, and give revenue allocation for the cloudfederation. However, creating a resource pool of coalition can incur a coalitional cost,in this case, we analyze the stability of the coalition formation. Finally, the developmentbottleneck of three cloud providers, Amazon, Rackspace and Microsoft, was analyzed,then we point out that the three cloud providers have a need to cooperate. The cloudproviders must make a revenue allocation mechanism when they form a coalition, we canpresent a suitable revenue allocation with the method of cooperative game. We use somenumerical examples to study the cooperative behavior of multiple cloud providers. Bycalculating, we find that the cloud providers can get higher revenue in the ferderationthan on his own, this provides theoretical basis for the cooperation of cloud providers. Analyzing different revenue distribution scenarios may have certain reference to makinga suitable revenue allocation for the cloud providers.