Leveraging Slack Capacity in IaaS Contract Cloud Services

Abstract

This paper develops pricing and capacity planning models for a cloud provider taking into account the trade-off between service reliability and capacity utilization. The provider's objectives are to optimally allocate resources, price different types of cloud services, and ultimately extract the maximum revenue from the market. We first analyze two basic types of cloud services: contract and on-demand. We then propose hybrid models in which the two basic services simultaneously exist. For each hybrid model, we determine an optimal capacity planning strategy, optimal reservation fee, optimal per-usage fee, and corresponding service reliability. We conduct experiments to study how optimal solutions change with respect to critical system parameters such as service rate, market size, and service reliability. Our work provides a framework for resource allocation, service provision, and strategic planning in cloud computing practice.