An economic approach to networked computing with priority classes

Abstract
The operation of a networked computing system (NCS), such as the Internet, can be viewed as a resource allocation problem and can be analyzed using the techniques of mathematical economics. We define a general NCS and translate that set‐up into a model of an economy. The preferences of users are taken as primitives, and servers (providing database services including entertainment and news or computational services) in the network are viewed as productive firms with priority input queues. Each server charges a rental price for its services by priority class. We characterize optimal system allocations and derive formulas for supporting rental prices and priority premiums, such that the aggregated individual user demands do not exceed optimal levels, and waiting‐time expectations are correct. We propose and implement a decentralized price adjustment process. Some results from a simulation study are presented and discussed. Profit measures for each server can be used to guide investment decisions.