Abstract. This paper discusses the rationales for a Grid market and, in particular, the introduction of a market place for trading commoditized computing resources. The market place proposed makes computing resources from different providers substitutable through virtualization. This includes the definition of a spot and future market as well as the parameters that a market mechanism for computing resources should consider. The above market place is complemented by a set of value-added services (e.g. insurance against resource failures, capacity planning, resource quality assurance, stable price offering) that ensure quality for Grid users over time. The market place technology for all of the above services has been designed by the GridEcon project, contributing to a broader adoption of Grid technology and enabling a service-oriented knowledge utility environment.
Commercial Grid markets have been a topic of research for many years. Many claims about the advantages of trading computing resources on markets have been made. However, due to a lack of Grid computing offerings, these claims could not be verified. This paper analyzes the question whether using the Grid is financially advantageous, using the Amazon.com EC2 service as a reference. To perform this analysis, the costs of computing resources in different usage scenarios are calculated, if Grid resources and in-house resources are used. The comparison of the costs reveals that while the Grid is cheaper in the short term, it is not a good investment in the long term and, thus, the existence of a Grid economy will not lead to an end of ownership but rather to a reduction in in-house resources and more efficient resource usage.
Research into computing resource markets has mainly considered the allocative fairness of market mechanisms. It has not been discussed how a large variety of resource types influences the market liquidity. Markets containing large numbers of buyers and sellers for heterogeneous resources suffer from a low likelihood of matching offers and requests. Traders therefore have the high risk of not being able to trade resources. We suggest a solution that derives SLA templates from a large number of heterogeneous SLAs in the market and, by using these templates instead of the original SLAs, facilitates SLA mapping. The usefulness of this approach is demonstrated through simulation results and a comparison with an alternative approach, in which SLAs are predefined.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.