Tender based resource scheduling in grid with Ricardo's model of rents

Abstract

The policy of resource scheduling determines the way in which the consumer jobs are allocated to resources and how each resource queues those jobs such that the job is finished within the deadline. Tender based scheduling policy requires the resources to bid for job while the consumer processes the bid and awards the job to the lowest bidder. Using an incentive based approach this tender based policy can be used to provide fairness in profit for the resources and successful job execution for the consumers. This involves adjusting the price, Competition Degree (CD) and the job position in the resource queue. In this paper, this model is further extended by incorporating resource categorization and modifying the resource bidding using 'Group Targeting'. The resources are classified using the 'Ricardo's theory of rents' taking into account the speed and type of each resource. This allows the consumer to make his decision using the category of resource along with its price which induces a quality element in the bid processing mechanism. This is modeled using the parameter Quality Degree (QD) introduced in the consumer side. This categorization and modified bid processing result in a scheduling policy closer to market-like behavior. © 2011 Springer-Verlag.