A study of hierarchical cloud resource pricing

In the current IaaS cloud market, to achieve profit maximization, the cloud provider offers volume discount and congestion pricing, where cloud brokers dynamically aggregate traffic from tenant consumers using stochastic multiplexing techniques. At the same time, tenant consumers judiciously adjust demands when reserving resources from brokers. Specifically, an interrelated market is formed, where brokers procure resources from the cloud provider and then sell the resources to tenant consumers. In this paper, we propose a practical hierarchical resource pricing model to investigate strategic interactions among tenants, brokers, and the cloud provider in both competitive and oligopoly market scenarios. Optimal demand response of tenants and its impact are scrutinized. We then extend our model to the case where tenants may have delay-tolerant traffic. Our evaluation is conducted using data from Google cluster traces, and reveals insightful observations for both theoretical analysis and practical pricing scheme design.