E-science Workflow on the Grid

Grid computing, which can be characterized as large-scale distributed resource sharing and cooperation, has quickly become a mainstream technology in distributed computing. In this paper, we present the idea of applying certain grid workflow management techniques to mediate various services for grid-based e-science processes. The techniques of adaptable workflow services, aggressive sub-workflow scheduling, and application-level checkpointing are used to improve the performance and trustworthiness of e-science workflow on the grid. A protein structure prediction application is presented as an example to show how the embedding of a large-scale e-science application in a grid-based workflow can be used to achieve high performance simulation in this context.