Bridging VisTrails Scientific Workflow Management System to High Performance Computing

Abstract

NASA Earth Exchange (NEX) is a collaboration platform whose goal is to accelerate Earth science research, by leveraging NASA's vast collections of global satellite data together with access to NASA's High-End Computing (HEC) facilities. NEX also aims to facilitate the sharing of experimental results as well as scientific processes (workflows) with the Earth science community through integration with VisTrails workflow management system. While VisTrails is used internally, it is not easily accessible from remote computers without directly logging into the NASA HEC systems through twofactor authentication and a bastion host. This paper describes the initial design of an extensible architecture that facilitates easier workflow interaction on NEX, by enabling users to develop and execute workflows in a supercomputing environment directly from their local VisTrails installation. This architecture helps domain scientists seamlessly leverage distributed computing and storage resources and it is potentially applicable to other scientific workflow management software. We further describe the architecture of the VisTrails-HEC plugin (as well as the VisTrails-Amazon plugin) and the implementation of a working prototype to demonstrate the feasibility of our solution.