Towards Simplifying Workflow Execution on The Grid
Researchers often spend valuable time performing mundane tasks to prepare for processing data on HPC. This is a function of two problems: first, researchers are forced to use unfamiliar tools to perform tasks on a compute cluster; and second, typical data management practices are not conducive to quickly moving data to a compute cluster for processing. In this paper, the authors propose a system that solves both these problems. Using VisTrails in conjunction with the Globus Toolkit, researchers can use simple visual tools to submit jobs to a variety of clusters, effectively making scheduling tasks transparent (i.e., the researcher is not required to interact with any particular schedule).