Unified Process for e-Science (UPeSC)

1
Unified Process for e-Science(UPeSC)

• NeSC
• 12-13 December 2002

2
USDP History

• 1967 Ericsson components
• 1987 Jacobson Objectory Process
• 1995 Unified Method
• 1996 Rational Objectory Process
• 1997 Three Amigos? Unified Modeling Language
• 1998 Rational Unified Process and still going

3
Core features

• Use-Case Driven
• Architecture-Centric
• Iterative and Incremental
• Organisation Phases Workflows

4
Use-Case Driven

• Capture functional requirements
• gives a user a result of value
• Drives development process
• Design
• Build
• Test

5
Architecture-Centric

• common vision
• Most important model elements
• Different views

? Architectural baseline (small skinny system)
6
Iterative and Incremental

• Iterations mini-projects
• Use-case selection
• Most important risks
• Incremental development

7
Extensions

• RUP Rational Unified Process
• EUP Enterprise Unified Process
• UPeSC Unified Process for e-Science

8
RUP

• New workflows
• Business modeling
• Deployment
• Config Change Mgt
• Project Mgt
• Environment

9
EUP

• RUP workflows
• Ops Support
• Enterprise Mgt
• RUP phases
• Production
• Retirement

10
UPeSC Evolution (by accident, not design)

• Use-cases
• Science problems
• Sequence diagrams
• Domain model
• Use-cases

11
Science Problems

• Focus on problems to be solved
• What is difficult?
• What takes too long?
• What is exciting?
• ?

12
Sequence diagrams
13
Domain model
14
Hindsight

• Start with Science
• Work packages contra UP spirit
• Many concerns vs ONE
• Domain model not especially useful

15
What is/was missing?

• Hire Software Architect from (before) start
• ? VISION
• Case studies in UP for e-Science/academic
  projects
• Iterative planning process
• Metrics
• milestones/deliverables setting
• Methods of collaborative working(in distributed
  team)

16
What is good!

• Acceptance of UP
• But still insisting on wp-based metrics
• Risk-willing (vs risk-averse)
• Academics are used to
• Incremental development
• Refactoring

17
Next steps

• Write-up AstroGrid as case study
• Investigate planning approaches
• (system dynamics approach?)
• Better marriage of work package structure
  single deliverable
• Develop/promote believable metrics

18
References

• http//www.rational.com/products/rup/
• http//www.therationaledge.com
• http//www.ambysoft.com/
• http//www.ronin-intl.com/publications/unifiedProc
  ess.html