Software Engineering in Action

1
Software Engineering in Action
Configuration Management Systems
Atsushi Inoue
Computer Science Department Eastern Washington
University
2
In This Talk...

• Share My Experience on a Large Scale Software
  System Development
• How Important Software Engineering Is
• Especially, Requirement Captures

3
What Project Was I Involved?

• Configuration Management System (CMS) for Nuclear
  Electric Power Plants
• I was a Researcher of CAD/CAM/CAE
• Research Development of Near-Future Products
• Consulting
• Patent Management
• Participated as a SE Consultant
• 1991, a year after I graduated

4
Configuration Management Systems

• Information System containing ALL information to
  support a particular product life-cycle (target
  product nuclear power plants)
• Architecture a large-scale, distributed database
  system with variety of API
• Effectiveness reduce "indirect" cost QA
  (ISO-9000)
• Functionality provide data access from many
  different aspects

5
How Large (and Complex)?

• Nuclear Power Plant 10 years of product span
  (from order to completion)
• 3000-5000 people involved
• Total Development Cost Several 100M / site
• 6000-8000 A1-size pages of diagrams
• Multiple companies (Electric, Construction, etc.)
• Assessment by government (different aspects from
  design and analysis)

6
What was Happening?

• On-going project, getting behind the schedule
• 1 year already spent on analysis
• supposed to be in design phase
• releasing the first functional prototype
• Experienced project team (5) in IT Department
• Loosing competence by facing new technology OO,
  EWS/PC (downsizing), RDBMS (ORACLE, ADABAS)
• Great amount of miscommunication --gt lack of
  trust between the project team their clients

7
Who are Involved?
Project
Straighten Up
8
What did I do?

• Reduce Miscommunication
• I wrote MANY scenarios (similar to play scripts)
  of using the system and its behaviors
• Use them as the primary document to address
  issues and problems in BOTH clients and IT
• Development of Prototype
• Concurrent Prototyping Approach (next slides)
• Promotion of Quality Control Working Group
  training new technologies -- UNIX, OOP, OOA/D, SQL

9
Concurrent Prototyping Approach

• Simultaneous Requirement "Re-"capture and
  Prototype Development
• Me Write/Revise Scenarios
• Project Members Design, Programming, Testing
• Standardizing Diagrams/Documents
• Promoting use of standard diagrams ERD, SC
• Use "Pseudo" Objective-C like class definitions
  to specify interfaces
• Key Version/Revision Management

10
From Software Engineering Aspects

• Impacts of Requirements Captures (hard way?)
• I thought it was easy, but it was the hardest
  part.
• Aspects of Rational Unified Process
• Use-Case Driven (I used scenarios use-cases)
• Architecture Centric
• Iterative and Incremental by being risk-driven
  (requirement risks and skill risks)
• This was prior to the birth of UML/RUP (1994)

11
What I Learned

• Communication (project team), communication
  (clients) and communication (executives) they
  are very important
• Power of Object-Oriented Technology in Many
  Aspects (Analysis, Design, Programming)
• Effectiveness of Concurrent Approach "attitude"
  of being proactive
• Later Power of Formal Methodology UML/RUP