May%202005

1

• May 2005

Greg Giacovelli Nick Mancuso Shaun Newsum
Jean-Paul Pietraru Nick Stroh
2
Agenda

• Background
• Motivation
• Requirements
• Schedule
• Metrics
• Architecture
• Tier Design
• Domain experience

3
Background

• Replace the try system
• Compiles, tests, and saves students submissions
• Provides immediate feedback to the student
• Collection of scripts written in the early 1990s
• Issues
• Steep learning curve
• Performance issues at peak hours

4
Motivation

• New Features
• Cross platform support
• Small learning curve
• A solid framework
• Provide clear ports for adding future
  customizations expansions
• Separate concerns to ease maintenance
• Distribute CPU load

5
Requirements

• Gathered from
• Project sponsor
• Personal experience with old system
• Sample try lab configuration
• Tested against
• Project sponsor
• try manuals
• SRS is thorough will be useful to future
  developers

6
Schedule
Due Date Delivered Name Description
12-26-04 12-26-04 Project Plan Plan and schedule
01-06-05 01-26-05 SRS Defines the system requirements
01-25-05 02-18-05 Design Describes the system design
01-20-05 01-20-05 Test Plan Defines the testing approach
01-29-05 02-18-05 M1 Milestone 1 The Framework
02-17-05 02-17-05 Progress report Midpoint project presentation
02-28-05 03-24-05 M2 Milestone 2 Submit
03-31-05 05-02-05 M3 Milestone 3 Compile and Test
04-28-05 04-28-05 Project poster Product presentation poster
04-30-05 N/A M4 Milestone 4 Code complete
05-16-05 05-16-05 Conference Paper End of term paper
05-20-05 N/A M5 Milestone 5 Gold
7
Metrics
Man Hours (Est.) 660
LOC 23550
LOC / Man Hour 35.6
8
Metrics
9
Solution

• Architectural framework
• May be easily extended to meet all requirements
• Runs in any J2EE-compliant application server.
• Scalable by distributing amongst multiple hosts
• Distributed submission processing
• Compile testing is completed by an array of
  clients
• Increases system performance
• Scales up by adding more client machines

10
Solution

• Model-View-Controller
• Decouples
• UI from business logic
• Business logic from data
• Web services
• Interface to the business logic
• Custom UIs scripts may be written

11
Architecture
12
Web Tier
13
Business Tier
14
Data Tier
15
Middle Tier
16
Problems we encountered

• Configuration issues
• jBoss
• Integration tests
• J2EE complexity
• Learning curve

17
Post Mortem

• What went well
• Solid documentation throughout.
• Proof-of-concept or better code for each major
  component of the system.
• What we learned from
• Too much wind-up time
• Should have dropped JBoss early

18
questions? comments?concerns?