Essays on sharing computer and information resources

Sharad Barkataki, Purdue University

Abstract

In this dissertation we study problems associated with the sharing of computer and information resources. The first two essays examine optimal pricing mechanisms in Distributed Computing Systems from a theoretical and experimental perspective. We build a theoretical model of a computing grid with revenue sharing and show that free and flat pricing regimes are suboptimal. We then find the optimal pricing mechanism and extend it to the case of multiple job classes. In our second essay, we conduct an experiment on a similar DCS to the one described in chapter 2. We additionally examine the impact of group size and congestion level on the performance of three mechanisms: a free regime, a fixed price regime, and a marginal congestion pricing regime. Our results indicate that in low congestion small user settings, a flat pricing mechanism actually performs better in terms of generating social efficiency than the marginal congestion pricing scheme. However, in a high congestion setting with a larger number of users, we find that the marginal congestion mechanism appears to do a little better than the flat pricing regime. Our final essay examines problems associated with information sharing in supply chains. We study the impact of introducing a Vendor Managed Inventory into a supply chain with a single retailer and multiple suppliers in the presence of product substitution. We first build a model of demand substitution and then use it to analyze the impact of VM1 introduction in the presence of adoption costs and variable stocking costs. We illustrate that prior solutions to this problem suggesting that stocking quantities will always increase under the Vendor Managed Inventory system to be incomplete. We show that the solution depends significantly on the underlying cost parameters and may vary accordingly. We also illustrate the conditions under which retailers and suppliers will benefit or lose from the introduction of the system, and identify optimal transfer payments that retailer may be willing to make to encourage adoption of such systems.

Degree

Ph.D.

Advisors

Chaturvedi, Purdue University.

Subject Area

Management|Systems design

Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server
.

Share

COinS