Declarative task delegation in OpenMOLE

In this paper we present OpenMOLE, a scientific framework providing a virtualized runtime environment for distributed computing. Current distributed execution systems do not hide the hardware and software heterogeneity of computing and data resources whereas OpenMOLE provides generic services to develop distributed scientific algorithms independently from the execution environment architecture. OpenMOLE uses abstraction layers to delegate computing tasks with the same high level interface for the major underlying architectures: local processors, batch systems, computational grids, Internet computing and cloud computing. The file access abstraction layer is another key feature helping a generic usage of the computation power provided by grids and clusters. The OpenMOLE framework has been tested with the exploration of a bacterial biofilm simulation with an individual-based model.