A Framework for Applying Inventory Control to Capacity Management for Utility Computing

A key concern in utility computing is managing capacity so that Application Service Providers (ASPs) and Computing Utilities (CUs) address their service level agreements. We observe that many of the capacity management problems faced by ASPs and CUs are similar to inventory problems addressed by retailers and warehouses, an area where inventory control is widely used. Applying inventory control to capacity management has appeal since inventory control has a rich set of analysis techniques and is widely used in practice. This paper proposes a framework for applying inventory control to capacity management for utility computing. The framework consists of: conceptual foundations (e.g., establishing connections between concepts in utility computing and those in inventory control); problem formulations (e.g., what factors should be considered and how they affect computational complexity); and QoS forecasting, which is predicting the future effect on QoS of ASP and CU actions taken in the current period (a critical consideration in searching the space of possible solutions).