Objective: study how to build e-Science applications and their supporting orchestration environments (workflows, bags-of-tasks, many-task computing environments) based on a consistent set of middleware- independent programming interfaces (API's).
Background: The e-Science community is divided in different middleware camps, focusing on and being bound to their respective middleware environment, for example, gLite, Globus, NorduGrid, Unicore, OMII- UK, or cloud systems like Amazon EC2, Nimbus, Eucalyptus, or OpenNebula. These (and other) middleware platforms provide their native programming models and interfaces; there is no consensus or coordination among these interfaces. Applications and support environments are thus being developed for these particular middleware systems, prohibiting the use of these community-specific codes across different middleware or even middleware releases.
Description: Develop a consistent and cohesive system of API's that supports both e-Science applications themselves as well as supportive programming environments like workflow systems or, more generally, many-task computing systems. Starting point will be the OGF-standardized Simple API for Grid Applications (SAGA) and Open Cloud Computing Interface (OCCI). Build a system of components that implement these API's and facilitate the orchestration of e-Science applications with these components. Evaluate (and possibly refine) these API's using e-Science applications and workflow systems used within the project.