Scalable Grid Resource Allocation for Scientific Workflows Using Hybrid Metaheuristics

Grid infrastructure is a valuable tool for scientific users, but it is characterized by a high level of complexity which makes it difficult for them to quantify their requirements and allocate resources. In this paper, we show that resource trading is a viable and scalable approach for scientific users to consume resources. We propose the use of Grid resource bundles to specify supply and demand combined with a hybrid metaheuristic method to determine the allocation of resources in a market-based approach. We evaluate this through the application domain of scientific workflow execution on the Grid.