【Title】Pricing and Capacity Management in the Cloud Computing Industry
【Speakers】
1. Kamran Moinzadeh, Chair of the Information Systems and Operations Management,Michael G. Foster Endowed Professor of Operations Management,Faculty Co-Director of the Master of Supply Chain Management Program,Foster School of Business, University of Washington at Seattle;
2. Shi Chen, Assistant Professor of Operations Management, University of Washington at Seattle.
【Host】Dr. Tianhu Deng
【Time】14:00–15:30, March 30th, 2018
【Location】 Room N412 , Shunde Building

【Abstract】Cloud computing has been recognized as one of the rising trends in the business world, and the recent surge in demand for cloud services has posed challenging problems for the cloud providers, especially their pricing and capacity management strategies. This presentation is based on findings
in two of our most recent research papers.
 
In the first paper, we study two important pricing schemes offered by major service providers in the cloud industry: the reservation-based scheme (namely the R-scheme) by Amazon and Microsoft, and the utilization-based scheme (namely the U-scheme) by Google. We develop a duopoly model with heterogeneous customers characterized by the mean and the coefficient of variation of their usage. We show that the effective price of either scheme is an increasing function of the coefficient of variation of the customer usage and find that when the providers use different schemes, customers with small demand volatility would prefer the R-scheme and those with large demand volatility would prefer the U-scheme. Furthermore, we examine the impact of evolving market characteristics, such as market preference, customer size, and market volatility, on the service providers' choices of schemes, settings of the pricing parameters, and the social welfare.
 
In the second paper, we aim to tackle the following problem: While the growths of demand for capacity attributes (e.g., CPU and RAM) are time-varying and disproportionate, replenishments of these attributes are often in pre-configured packages (e.g., server clusters); the ratio of supply in a specific package does not match with the ratio of demand of the attributes. We consider demand growths of two attributes and focus on a class of policies consisting of capacity expansion cycles, where capacities are added through sequential replenishments of two given cluster-types and excess capacities of the attributes are required to reach a desired minimum level at the start and the end of each cycle. For the linear demand growths, we identify the optimal policy in closedform; for the exponential demand growths, we devise a dynamic-programming (DP) algorithm, as well as a forward-looking heuristic based on minimization of the total cost rate of each cycle. We also examine the problem of cluster selection, either static or adaptive, given a set of cluster-types.Furthermore, we numerically study the performance of the proposed policies: The Forward-Looking Heuristic policy and one of its special cases, the Minimum-Cycle-Length policy, perform very well, and both policies outperform a policy used in practice, the Single-Cycle policy; we also find that adopting an adaptive cluster-selection strategy could bring substantial cost savings, compared to the static strategy, especially when the discrepancies between the demand growth rates and the holding cost rates of the attributes are large.
 
All interested are welcome!