Pricing Schemes in Cloud Computing: Utilization-Based versus Reservation-Based

Abstract

Cloud computing has been a rising trend in the business world. In this paper, we consider two most important pricing schemes offered to sustained customers by major service providers in the cloud industry: the reservation-based scheme (the R-scheme) by Amazon or Microsoft, and the utilization-based scheme (the U-scheme) by Google. We consider a duopoly model with heterogeneous customers characterized by the mean and the coefficient of variation of their usage. We show that under either pricing scheme, the effective price is essentially an increasing function of the coefficient of variation of usage, and thus both schemes aim for rewarding stability in usage. However, when the providers adopt different schemes, we show that customers with lower demand volatility would prefer the R-scheme, while those with higher demand volatility would prefer the U-scheme. Furthermore, we study the impact of evolving market characteristics, including the distributions of market preference, demand size, and demand volatility, as well as the impact of the providers' service levels on their choices of schemes and decisions on the pricing parameters. We find that if the market has a stronger preference for a particular provider or that provider has a higher service level than its competitor, the provider is more likely to adopt the R-scheme, while its competitor's adoption of a scheme depends on the extent of the price competition. Specifically, when the diversity of customer preference becomes higher (lower), the price competition becomes softened (intensified), and the competitor is more likely to adopt the R-scheme (U-scheme, respectively).

Keywords: Pricing Schemes, Sustained Use Discounts, Reserved Instances, Capacity Planning, Cloud Computing