Cocomo Lecture III EEE493 2000

1
Cocomo - Lecture III EEE493 2000
Royal Military College of Canada Electrical and
Computer Engineering

• Major Greg Phillips
• greg.phillips_at_rmc.ca
• 1-613-541-6000 ext. 6190

Dr. Scott Knight knight-s_at_rmc.ca 1-613-541-6000
ext. 6190
2
Refs

• Pressman, R.S., Software Engineering a
  Practitioners Approach 5th Ed., McGraw-Hill,
  2001, Chapter 4
• Boehm, Barry, Software Engineering Economics,
  Prentice-Hall, 1981

3
Teaching Points

• Maintenance using the Intermediate Model
• Sensitivity/options analysis

4
Review

• What are the three modes of development under
  Basic COCOMO?
• How does Intermediate COCOMO differ from basic
  COCOMO ?
• Since we are using a product of the individual
  cost driver effort multipliers as the adjustment
  factor, what must be a basic assumption about the
  cost drivers?

5
Recall A Pricing ExampleMicroprocessor
Communications Software
6
Cost Driver Ratings Microprocessor
Communications Software
7
Project Cost Microprocessor Communications
Software

• PMnom 2.8(10)1.2 44 pm
• PMadj 44 pm ? 1.17 51 pm
• 51 pm ? 6000 /pm 306,000

nominal
adjusted
8
(No Transcript)
9
What if we used less expensive personnel?

• I.e. less capable analysts and programmers
• Remuneration is now 5000/month
• Therefore ACAP and PCAP are now nominal in value
• The effort multipliers are 1.00 vice .86
• Effort Adjustment Factor is now 1.58
• PMadj 44 pm ? 1.58 70 pm
• 70 pm ? 5000 /pm 350,000
• Therefore senior analysts and programmers are
  more economical (by 44,000) in this case

10
What if we bought more memory?

• For 10,000 Megabit could by 96K words of main
  memory vs. the current 64K
• Main storage constraint 45K/96K .47
• Therefore STOR is now nominal
• I.e. STOR 1.00
• Effort Adjustment Factor is now 1.10
• PMadj 44 pm ? 1.10 48 pm
• 48 pm ? 6000 /pm 288,000
• The 18,000 saving offsets the 10,000 expense

11
Options Analysis

• We can use these sensitivity analysis techniques
  to cost different development scenarios
• This is an optimization problem
• We are trying to minimize the cost-to-complete a
  project
• But what about the lifecycle cost?

12
Adjusted Annual Maintenance

• Most cost drivers are the same for the
  maintenance phase as for the development phase
• The exceptions are
• SCED
• RELY
• MODP

13
SCED under maintenance

• SCED is always nominal
• development schedule does not effect maintenance

14
RELY under maintenance

• Consider
• high reliability -gt more effort to reach the
  required level of quality
• lower reliability -gt less effort to maintain the
  required level of quality
• lower reliability -gt more effort to fix the
  latent faults in the code, docs, etc.

15
MODP under maintenance

• The more MPPs (modern programming practices)
  used the greater the savings in maintenance
• The more MPPs user the greater the economies in
  maintaining larger projects (i.e. the more
  success in reducing the diseconomies of scale)

16
Maintenance Example

• Use the example as before of the 10 KDSI
  communications product with the 96K main memory
  upgrade option
• Assumptions
• product will be maintained by nominal programmers
  and analysts _at_ 5000/pm
• virtual machine volatility will be low vs.
  nominal
• virtual machine experience will be nominal vs.
  low
• annual traffic change 20
• nominal development effort was 44 pm

17
Effort Multipliers
18
Maintenance Example (contd)

• PMAM 1.14 ? .20 ? 44 pm 10.0 pm
• FSPAM 8.8 pm/12 months 0.83 fsp
• 10 pm ? 5000/pm 50,000 (yearly)

19
Lifecycle Cost

• Consider a 10 year lifecycle for our
  communications software
• Assume the example as before with the 96K main
  memory upgrade option
• Development
• 288,000
• Maintenance
• 50,000/year ? 10 years 500,000
• Lifecycle cost
• 788,000

20
Discussion

• What about the sensitivity to the primary cost
  driver?
• What about the relevance of the models database?
• Are the cost drivers likely to remain constant
  across the phases of the development process?

21
Next ClassConfiguration management