Cost-efficient Resource Allocation for Decentralized Clouds

Cloud computing is gaining strength in the IT marketplace because of its on-demand resource provisioning, high availability and elasticity. To offer these advantages, many largely centralized computing infrastructures have emerged, which results in many concerns such as resource over-provisioning, huge energy consumption, privacy protection, data security, etc. In this proposal, we envision a decentralized cloud in which the resources are provided and shared by a group of small data centers. Through cooperations among small cloud vendors, the decentralized cloud can overcome the limitations raised by centralization. To further explore this topic, we have selected three relevant papers that pave the way for: 1) resource heterogeneity and recommendation in a customer-provided resource cloud platform; 2) price and QoS competition games in cloud service markets; 3) resource sharing model in a federation of selfish cloud providers. Finally, based on the previous work, we discuss our research plan which mainly focuses on the cost and incentive models of the resource allocation.