RADICAL-Cybertools: An abstractions-based approach to Scientific Workflows

Presenters: Shantenu Jha, Associate Professor, Rutgers, The State University of New Jersey University; and Ioannis Paraskevakos, Research Assistant, Research in Advanced DIstributed Cyberinfrastructure and Applications Laboratory (RADICAL), Rutgers, The State University of New Jersey University

Date: May 10, 2017

Slides: https://uofi.box.com/s/85qym36n1xmiiad6rf7nibyhjmo3y7ws

Video: https://www.youtube.com/watch?v=1zuC4Wy8ynE&list=PLO8UWE9gZTlAMRvvVfS7-6q3x1DrXKmkR&index=10


RADICAL-Cybertools enable the scalable execution of simple and common workflows, while allowing the construction and extension of workflows and tools with specific and complex requirements. They are tools that provide interoperable access to compute and data resources, while supporting the concurrent execution of multiple tasks.

Each component of RADICAL-Cybertools provides a well defined capability, such as task-level parallelism. RADICAL-Cybertools can be used by domain scientists “as is”, or as part of a software stack that allows developers to build scripts/tools that meet specific and complex requirements.

The three main RADICAL-Cybertools are Ensemble-Toolkit, RADICAL-Pilot and RADICAL-SAGA, each one implementing well-defined abstraction.

Target audience: Computational scientists, students, tool developers.


  • basic knowledge of Python
  • basic knowledge of Unix command line
  • basic knowledge of HPC usage (batch systems)
  • basic knowledge of Spark (desired, not required)

User base: Computational scientists, HPC users

Software availability: Open Source under MIT License

Software requirements: Python 2.7

Use cases: Molecular Dynamics Analysis (e.g. Leaflet Finder)

Training and reference materials: http://radical-cybertools.github.com