An equilibrium pricing model for large scale computational markets

Abstract

Load balancing is one of the most challenging problems facing large-scale computational systems. This paper introduces a competitive pricing model to achieve a general equilibrium between the global supply and demand in large scale computational commodity markets. Prices are adjusted according to a tâtonnement like process. A fully distributed pricing model is proposed based on an existing partially distributed version. While in this last version, the price of each commodity is computed by only one auctioneer, in our algorithm, a scalable number of auctioneers is used to support a massive number of suppliers and consumers.