A SOA-based environment supporting collaborative experiments in e-Science

Many sophisticated environments allow creating and managing of scientific workflows, whereas the workflow itself is provided as a service. Scientific Grids handle large amounts of data and share resources, but the implementation of service-based applications that use scientific infrastructures remains a challenging task, due to the heterogeneity of Grid middleware and different programming models. This paper proposes an e-Science environment providing functionality in a simplified way, considering the Grid as a source of computational power and an information infrastructure. To promote integration among components and user interaction, the paper outlines a SOA-based scientific environment where an experiment is modeled through an abstract workflow defining the functional model of the experiment. The tasks are mapped to the corresponding scientific services by a workflow engine, separating logical aspects from implementation issues. Services depend on the type of experiment and can be re-used, wrapped, or moved into a new workflow. Infrastructural services discover suitable resources that match user requirements and schedule workflow tasks. Further, they monitor the execution of each task and aggregate the results. The proposed approach provides a simple-to-use and standardized way for the deployment of scientific workflows in a distributed scientific environment, including the Grid.