A service-oriented framework for integration of domain-specific data models in scientific workflows

Managing data exchange in scientific workflow simulations is a challenge for which existing solutions have only limited success. Basing on a data-model approach we developed a service-oriented, platform-and language-independent framework which provides seamless access for heterogeneous applications to data storage resources distributed over grids and clouds. After implementing a working prototype of the framework we demonstrated its benefits for a complex workflow application from the domain of multiscale materials modeling. Because of its generic architecture and wide standards conformity the framework can be deployed as grid or cloud middleware for various other domains of computational science to enable rapid development of complex workflow applications.