Essays on improving software development productivity and economic and operational analysis of content distribution networks.

摘要

The first essay on improving software development productivity addresses issues related to a new software development approach called pair programming. The remaining two essays in this dissertation focus on economic and operational issues in content distribution networks. Each essay uses methodologies such as integer programming, numerical simulation, and game theory to tackle these Information Systems issues.;Recently there has been considerable interest in pair programming---a novel development approach in which a pair of developers jointly works on the same piece of code. The purpose of the first study is to compare the performance of pair development with that of solo development under two separate objectives---effort minimization and time minimization. In both cases, we develop analytical models to calculate optimal module to developer assignments (for both the pair and solo approaches) so as to minimize the objective being served.;In the second essay, we develop an economic model to investigate the impact peer-to-peer (P2P) networks on the providers of content distribution services (such as Akamai). Our finding is that the eventual impact of a P2P network could either benefit or hurt the profits of a content distribution provider (CDP). Our model considers the incentives of the producer (or content provider, e.g., CNN) together with the incentives of the consumer (who derives some positive utility from consuming the content) to derive the optimal extent of content distribution, with and without the existence of a P2P network.;In the third essay, we present and analyze a simple, structured model of peer-to-peer (P2P) networks to gain insights. The primary objective of this study is to examine whether sharing in P2P networks can result purely from self-interest. Another goal is to understand the conditions under which free-riding can occur. We find that congested external common resources, such an external link through which all traffic enters and leaves the P2P community, play a key role in determining the sharing profile of a P2P community. Finally, free-riding can be shown to occur if the participants are heterogeneous with respect to the resources they possess.