The Workbench An Introduction to Scientific Workflows Robert Bridgart | Software Engineer September 2013
CSIRO LAND & WATER / WATER FOR A HEALTHY COUNTRY FLAGSHIP Workflow
2 | The Workbench: An Introduction to Scientific Workflows | Robert Bridgart Scientific Workflows
What is a scientific workflow?
“A scientific workflow system is a specialized form of a workflow management system designed specifically to compose and execute a series of computational or data manipulation steps, or a workflow, in a scientific application.”
http://en.wikipedia.org/wiki/Scientific_workflow_system
3 | The Workbench: An Introduction to Scientific Workflows | Robert Bridgart Project Trident A Scientific Workflow Workbench
Origin - Microsoft Research
Currently - Open Source project w/Outercurve Foundation
Microsoft .NET based implementation of a scientific workflow system
Uses Microsoft .NET Workflow Foundation
4 | The Workbench: An Introduction to Scientific Workflows | Robert Bridgart Project Trident A Scientific Workflow Workbench
Provides tools and capabilities to
• Create, store and execute workflow steps (aka Activities)
• Manage workflow/activity revisions
• Record execution history of workflows (what, who and when)
• Persist data products (inputs, intermediate results and or outputs)
Competing technologies – Taverna and Kepler (both Java based)
5 | The Workbench: An Introduction to Scientific Workflows | Robert Bridgart The Workbench
• A suite of tools developed to extended and increase the useability of Workflow Foundation based scientific workflows
• Provides a selection of generic Activities and data types • Activities for importing / transforming / reporting datasets • Activities for calibration / optimisation (using HPC) • Definitions for custom data types i.e. Time Series
• Activity Generator – used to facilitate the embedding of existing programming tools and models into Activities • Programming languages – R, Python and .NET • Surface water models – Source, IQQM and Realm • Groundwater models - MODFLOW
• Utilities – database backup/restore, activity/workflow packing and distribution.
6 | The Workbench: An Introduction to Scientific Workflows | Robert Bridgart Surface water modelling example
7 | The Workbench: An Introduction to Scientific Workflows | Robert Bridgart eReefs project High level workflow
Estuary Lagoon Source Visualisation Modelling Modelling (Catchments) Platform (SHOC) (SHOC)
8 | The Workbench: An Introduction to Scientific Workflows | Robert Bridgart Demonstration
9 | The Workbench: An Introduction to Scientific Workflows | Robert Bridgart 10 | The Workbench: An Introduction to Scientific Workflows | Robert Bridgart 11 | The Workbench: An Introduction to Scientific Workflows | Robert Bridgart 12 | The Workbench: An Introduction to Scientific Workflows | Robert Bridgart End-to-end calibration
13 | The Workbench: An Introduction to Scientific Workflows | Robert Bridgart Why should I adopt the Workbench?
• Provides a framework for capturing and managing workflows in a central location
• Captures the provenance of output data
• Defines common data types (to facilitate activity/model interaction)
• Comes with a suite of predefined activities
• Provides support for a range of existing technologies (i.e. R, Source, etc.)
• Good support for calibration, optimisation and parallelisation using HPC
• Easily create new activities
• Accountability
14 | The Workbench: An Introduction to Scientific Workflows | Robert Bridgart Thank you CSIRO Land & Water / Environmental Information Systems Robert Bridgart Software Engineer t +61 2 6246 5885 e [email protected] w www.csiro.au
CSIRO LAND & WATER / WATER FOR A HEALTHY COUNTRY FLAGSHIP Getting Started with Trident
Prerequisites
• Microsoft Windows Vista SP3, 7 or Server 2008 • Microsoft SQL Server 2008 (Express or Full Edition) • Microsoft .NET 4.0 Framework or later
Optional
• Silverlight 3.0 with IIS 7.0 • Microsoft Office Word w/Trident Document Add-in package
Project Trident: A Scientific Workflow Workbench http://tridentworkflow.codeplex.com/
16 | The Workbench: An Introduction to Scientific Workflows | Robert Bridgart