Reputation-Based Pricing for Grid Computing in E-Science
Arun Anandasivam, Dirk Neumann · Journal of the Association for Information Systems · 2008
One of the fundamental aspects for an efficient Grid usage is the optimization of resource allocation among the participants.However, this has not yet materialized.Each user is a self-interested participant trying to maximize his utility whereas the utility is not only determined by the fastest completion time, but on the prices as well.Future revenues are influenced by users' reputation.Reputation mechanisms help to build trust between loosely coupled and geographically distributed participants.Providers need an incentive to reduce selfish cancellation of jobs and privilege own jobs.In this paper we present a reputation-based pricing mechanism for a simple, but fair pricing of resources.In e-Science researchers do not appreciate idiosyncratic pricing strategies and policies.Their interest lies in doing research in an efficient manner.Consequently, in our mechanism the price is tightly coupled to the reputation of a site to guarantee fairness of pricing and facilitate price determination.Furthermore, the price is not the only parameter as completion time plays an important role, when deadlines have to be met.We provide a flexible utility and decision model for every participant and analyze the outcome of our reputation-based pricing system via simulation.