Swing Options: A Mechanism for Pricing Peak IT Demand

As information technology (IT) and services become more prevalent within the enterprise, the burden of greater expertise on computer technology falls on companies whose primary mission may not be information technology management. This has led to the emergence of computation resource services such as utility and on demand computing. We describe a pricing scheme for using various computational resources from an IT provider, called a swing option service. The resources, computons, may be construed as a combination of CPUs, disk storage, and network bandwidth. The pricing scheme relies on records of past resource usage to allocate and charge for IT services. The records may be specific to a particular user or may be aggregated over many users. The pricing method involves scheduling the future use of computational resources by a customer who buys swing options. Swing option contracts for computational resources consist of reservations for the right, but not the obligation, to use a IT resources at a particular time in the future. The swing options span a lower and upper range for the customer’s use. Customer’s pay no less than the lower swing range, but pay a penalty for going above their upper range. The essential idea behind swing contracts is that they incentivize predictable usage. Predictable usage benefits the resource provider because resource planning can be done ahead of time, which is less costly than providing IT demand on a short-term basis. Predictable usage also provides a higher degree of customer service satisfaction. Thus, customers also benefit from predictable usage because they can be assured that resources will be available when they are needed at an agreed upon price and don’t have the burden of maintaining them when they are not needed. We derive pricing models that parameterize what swing option prices should cost to the customer based on a desired profit margin for the resource provider. We present simulations of the swing option service using historical data that agree with the expected profits.