Workflows naturally arise in computational applications. In Earth sciences they are used to automate data downloads, all sorts of forecasts, and climate research, among many others. However, their employment in the research of non-computational scientists is limited.
This short course aims to introduce workflows to an audience with minimal computational knowledge. We will see how workflows are currently used in Earth sciences, what workflow managers are, and how they can improve productivity.
Then, we will have a demonstration by four major workflow managers used in Earth sciences:
- ecFlow (Marcos Bento, ECMWF)
ecFlow is a workflow manager developed at ECMWF that enables users to run a large number of programs (with dependencies on each other and on time) in a controlled environment. It provides tolerance for hardware and software failures, combined with restart capabilities.
ecFlow is used at ECMWF to run all of their operational suites across a range of platforms.
- Cylc (Oliver Sanders, UK Met Office)
Cylc (pronounced silk) is a general purpose workflow engine that also manages cycling systems very efficiently. It is used in production weather, climate, and environmental forecasting on HPC, but is not specialized to those domains. It is developed by the Earth Sciences New Zealand (formerly NIWA) institute.
Cylc is used to run the operational forecast of the UK Met Office, the Australian Bureau of Meteorology, and the Earth Sciences New Zealand institute.
- Sirocco (Matthieu Leclair, ETHZ)
Sirocco is a climate and weather workflow tool written in Python. It brings together characteristics of the AiiDA library by handling data as explicit graph nodes alongside tasks and by defining task plugins and characteristics of Cylc with date cycling. Sirocco was developed between ETHZ and the Paul Scherrer institute (PSI) where the ICON model plays a central role, so that, even though it’s not designed for it specifically, Sirocco ships with a user-friendly ICON plugin.
- Autosubmit (Bruno de Paula Kinoshita, Barcelona Supercomputing Center)
Autosubmit is a lightweight, self‑contained workflow manager tailored for climate research that combines an experiment manager, workflow orchestration, and monitoring in one application. It supports running jobs locally and remotely while tolerating networking and I/O errors. It also provides a GUI for external monitoring.
Autosubmit is currently used to run the workflow of the Destination Earth project.
Introduction of workflows and workflow managers to non-computational scientists
Convener:
Manuel Giménez de Castro MarcianiECSECS
|
Co-conveners:
Bruno De Paula KinoshitaECSECS,
Marcos Bento,
Matthieu Leclair,
Oliver Sanders