An online auction mechanism for cloud computing resource allocation and pricing based on user evaluation and cost

摘要

Auction-based virtual resource allocation is a major problem for cloud computing. However, previous studies have typically equated virtual resources with ordinary items and designs corresponding to specific auction mechanisms, which cannot truly reflect the characteristics of the virtual resource. This inability to reflect the virtual resource characteristics arises because the users have only the right to use resources without ownership, the use of virtual resources has significant impacts on the social welfare and revenue of the resource provider, and the virtual resource usage will generate operating costs. Considering the above issues, this work proposes a truthful online auction mechanism based on user evaluation and cost and applies it to the allocation and pricing of cloud computing virtual resources. Based on a pay-as-you-go model, users can submit multiple requirements at any time, but only one requirement can be satisfied, which is known as a multi-requirement, single-minded scenario. We prove that the resource providers can obtain increased social welfare and guarantee that the mechanism is truthful. For the resource allocation problem, we propose a heuristic algorithm to obtain the allocation result quickly and to maximize the social welfare of the cloud resource provider. The payment algorithm accounts for the interests of both the users and resource provider based on the dichotomy. We experimentally analyze the solution in terms of social welfare, execution time, resource utilization and users served.