Scientific workflow management systems and domain specific applications

1
Scientific workflow management systems and
domain specific applications

• Zhiming Zhao, Paul van Hooft, and Adam Belloum
• (The VL-e Integration team and SP2.5)

2
Outline

• Scientific workflow management systems (SWMS) and
  e-Science
• SWMS and scientific applications
• Solutions
• Activities

3
A big picture Scientific Workflow Management
System (SWMS)

• In an e-Science environment, a SWMS
• Automate experiment routines
• Rapid prototyping of experimental computing
  systems
• Hide integration details between resources
• Management of lifecycle of experiments
• A SWMS crosses different layers of middleware for
  data, computing tasks, information and knowledge.

4
People in workflow
5
Scientific Workflow Management System (SWMS)

• e-Science and e-Science environment
• Different layers of middleware for data,
  computing tasks, information and knowledge.
• SWMSs cross these layers
• A model,
• An engine,
• User support.

Knowledge
Information
Computing tasks
Data management
Generic Grid middleware
6
Serve domain specific applications

• Automate experiment routines
• Model experiment processes and automate them
• Rapid prototyping of experimental computing
  systems
• Utilise e-Science resources (software
  components, Grid services, and tools) to
  construct experimental systems
• Hide integration details between resources
• Provide metadata for resource integration
• Automate discovery and binding
• Management of lifecycle of experiments
• Manage meta data
• Runtime information.

7
Between SWMS and Apps.
Scientific experiments use cases and scenarios.
Application scientists or Domain experts.
Application process modeling.
Scientific workflow management systems.
E-Science SWMS scientists and engineers.
Rapid prototyping.
e-Science resources components, tools and
services.
Application engineers or e-Science engineers.
8
Solutions

• Short term solution
• Survey
• Recommendation
• Install in PoC
• Long term solution
• Abstract the features and incorporate in the VL-e
  framework.

9
Solutions

• Recommend existing SWMSs (e.g., VLAM-G, Taverna,
  Kepler and Triana)
• Conduct domain specific research.

Short term.
Long term.

• Learn lessons from the short term solutions.
• Abstract the generic features from the short
  solution SWMSs
• Make new design of the VL-e framework which
  enables the interoperability among domains

10
Activities

• Regular meetings
• Joint research paper writing
• Special issue Semantic Grid in Journal Web
  Semantics http//km.aifb.uni-karlsruhe.de/publica
  tions/semanticgrid/
• Understanding application use cases
• Understanding application resources interfaces
• Providing recommendations

11
Between SWMS and Apps.
Application scientists or Domain experts.
Application scenarios
E-Science applications
E-Science SWMS scientists and engineers.
Application engineers or e-Science engineers.
Resources tools, components, and services.