CS 562 Advanced SW Engineering

1
CS 562Advanced SW Engineering

• Lecture 5
• Tuesday, May 11, 2004

2
Class Format for Today

• Return Proposal 1
• Turn in Paper 2
• Questions
• Lecture 5
• Chapter 3 Application Examples

3
Questions?

• From last weeks class
• From the reading
• About the papers/project?
• Anything else?

4
COCOMO IIApplication Examples

• Boehm, et al
• Chapter 3

5
Overview

• 2 example systems
• Transaction Processing System (TPS)
• Airborne Radar System (ARS)
• Show how to use COCOMO II to
• Develop estimates
• Perform trade studies
• Justify purchase of SW tools
• Perform risk analysis, etc.

6
Transaction Processing System

• Client-server system designed to
• Interactively process queries
• Perform a variety of functions across a network
• Types of transactions could include
• Airline reservations, financial transactions
• Library searches, remote DB browsing, etc.
• TPS Software Capabilities
• Review Table 3.1, page 85

7
SW Dev. Organization

• Experienced client-server programmers
• Process improvement efforts underway
• Issues
• Technical 1st-level management
• Language (C vs. Java)
• Optimistic / pessimistic estimates
• Lines of code vs. function points

8
Estimation Process

• Estimate effort for client-side software
• Review process illustrated in Figure 3.2, p. 88
• Metrics
• Sizing figures in Table 3.2, p. 88
• Work breakdown structure (WBS)
• Figure 3.3, p. 89
• COCOMO Model scope
• Table 3.3, p. 90

9
Sizing System SW Component

• Need to distinguish between 4 categories
• New, Modified, Reused, COTS
• Convert reused SW into equivalent new LOC
• Review formulas on pages 90 91
• What is the bottom line estimate for the System
  component?
• Do the numbers make sense?

10
Sizing GUI Component

• What information is taken from table 3.2?
• What else needs to be considered?
• What is the bottom line estimate for this
  component?
• Do the numbers make sense?

11
Sizing Fault Diag. Component

• Start with estimate in Function Points
• How to convert to LOC?
• Initial estimate seems unreasonable
• How does one arrive at that conclusion?
• What needs to be done about it?
• What is the bottom line estimate for this
  component?
• Do the numbers make sense?

12
Estimate Effort using WBS

• 5 tasks outlined in Table 3.4, p. 93
• Develop software requirements (3.2.4.2.1)
• Develop software (3.2.4.2.2)
• Perform task management (3.2.4.2.3)
• Maintain configuration control (3.2.4.2.4)
• Perform software quality assurance (3.2.4.2.5)
• Review issues and considerations for each
• How are the numbers derived?

13
Develop COCOMO II Estimate

• Process is outlined in Figure 3.4, p. 96
• What assumptions are made at the beginning?
• Table 3.5, p. 97 shows scale factor ratings
• Tables 3.6 3.9 show cost driver ratings
• What assumptions are present in the rationale?
• Do the assumptions seem reasonable?
• Why / why not?
• Review COCOMO output in Figures 3.5, 3.6, p. 101
  nominal vs. schedule constrained

14
Compare Estimates

• What are the differences between WBS COCOMO II
  estimates?
• Why are they different?
• Which one is correct?
• How to resolve differences
• What approach does the book take?
• Do you agree with the approach?
• Why create 2 different estimates?

15
Bounding Risk

• Table 3.10, page 104 shows Top 10 high- risk
  areas for TPS
• How to evaluate prioritize?
• Discuss the various risks
• Review sections 1-10 on pages 103 107
• Do the suggestions and approaches for handling
  risk make sense?

16
Trade Studies

• COTS Software, Language, and Paradigm
• Review Tables 3.11 3.13, pages 108-111
• What approaches are taken for these studies?
• How do you interpret the results?
• Do you agree with the conclusions?

17
Assessing Life Cycle Costs

• Review formulas on page 113
• How do they differ for maintenance as opposed to
  development?
• Review Table 3.15 (p. 114) Figure 3.8 (115)
• What happens to cost/benefit analysis of COTS
  when life cycle is considered?

18
Airborne Radar System

• Complex, real-time embedded system
• Involves concurrent dev. of new hardware
• Upgrade to an existing system already in use
• Constructed using substantial reuse automatic
  translation
• 3 COCOMO models
• Application Composition, Early Design and
  Post-Architecture

19
ARS Description

• Hardware
• Central on-board computer
• Radar other sensor units
• Proprietary display console w/ input devices
• Trackball, programmable touch panel, function
  keys
• Software
• See Figure 3.9, p. 117 Table 3.16, p. 118

20
Prototype Analysis

• Estimated using Application Points
• See Table 3.17, p. 119 and Table 3.18, p. 120
• Learn more about Applications Composition model
  in Chapter 5
• What variables are considered in cost estimate?
• See Formula on page 117
• Do the numbers seem reasonable?

21
Early Design Model

• What is the Breadboard System?
• What can/cannot be considered at this stage?
• Review Scale Factors Cost Drivers
• Table 3.19, p. 121 and Table 3.20, p. 122
• Review Size calculations Table 3.21, p. 123
• What assumptions are made?
• What is the bottom line time estimate?
• Do the numbers make sense?

22
Full Development Top Level

• Use Breadboard system for additional input
• Productivity of automatic translation
• Quantify sizes of objects in lines of code
• Historical data for ratings profile (Figure 3.11)
• New Scale Factors Cost Drivers
• See Tables 3.22 and 3.23, page 128
• New Size calculations for full ARS Dev.
• See Table 3.24, page 130 and Formulas, page 131

23
Detailed Component Estimates

• Scale factors remain the same, but detailed
  cost-driver ratings and rationale differ
• See tables 3.25 through 3.30 (pages 133, 134)
• From detail, get new estimated time/effort
• See formulas on page 132 and 134
• Review COCOMO output, Fig. 3.14, p. 135
• How does the top-level estimate compare to the
  one with more detail?

24
Incremental Development

• 3 time-phased increments of individually
  delivered software builds
• Reqs prelim design done first for all
  increments
• Impl contains some work done in parallel
• Need to decide what to deliver when
• See bullet list on p. 136, phased plan on pages
  136-137
• Take increment volatility into account
• Review Figures 3.15-3.17 pages 138-139

25
For Next Time

• Read Chapter 4 in Boehm, et al
• Proposal 2 for Paper 4 due