Resource management and risk mitigation in online storage grids

This dissertation examines the economic value of online storage resources that could be traded and shared as potential commodities and the consequential investments and deployment of such resources. The value proposition of emergent business models such as Akamai and Amazon S3 in online storage grids is capacity provision and content delivery at acceptable levels of Quality of Services (QoS) to meet the increasing demand on timely supply of data-intensive services. These models essentially entail two critical managerial functions: resource management and risk mitigation under demand uncertainty over both locations and periods. These functions can be undertaken in both layers - that where multiple grid nodes cooperatively meet their clients' demand and that where a grid provider directly sells to its clients. This dissertation consists of five essays – Chapters 2-6. Following the same principle of Akamai, in the layer among the grid nodes, Chapters 2 and 3 model a Capacity Provision Network where node providers possess resource infrastructures and leverage their topographies under stochastic demand over locations. Based on the model of Amazon S3, that of a single grid provider and its clients, Chapter 4 examines the effects on the provider's risk and return of providing a spot market with dynamic prices to attract buyers and a forward contract to hedge against demand uncertainty over time periods, and Chapter 5 examines how and whether options could supplement such a pricing mechanism. Continuing with Amazon S3's example, in Chapter 6 we model the two layers in an integrated manner using bi-level programming and develop a pricing mechanism for the provider to maximize its revenue and for the buyer to efficiently allocate resources among its nodes. We analyze the risk of information life cycle management underlying such a two-layer system with a storage grid of multiple tiers where the data are connected with each other.