FG4 Friday, June 6, 2003 10:30 AM S OFTWARE D EVELOPMENT ON I - - PDF document

BIO PRESENTATION International Conference On Software Management & Applications of Software Measurement June 2-6, 2003 San Jose, CA USA

FG4

Friday, June 6, 2003 10:30 AM

SOFTWARE DEVELOPMENT ON INTERNET TIME-FASTER, CHEAPER, WORSE?

Girish Seshagiri Advanced Information Services

Girish V. Seshagiri

Girish Seshagiri is CEO of Advanced Information Services Inc., headquartered in Peoria, Illinois, USA. The AIS Development group is a winner of the 1999 SEI/IEEE Computer Society Software Process Achievement Award. Girish has a strong commitment to sharing of knowledge and ideas with the rest of the software community. Girish has addressed SEPG conferences in the U.S, Europe, and India as well as other conferences such as the International Conference on Software Process Improvement (ICSPI), COMPSAC, and SM/ASM. He has given invited talks on Software Process Improvement in Bilbao, Beijing, and Seoul. He provided the leadership to establish The Watts Humphrey Software Quality Institute (SQI) in Chennai, India. Girish is one of the co-founders of the Chicago Software Process Improvement Network (SPIN), Heartland SPIN, and the Chennai SPIN. Girish serves on the IEEE Software Industrial Advisory Board.

IT on Internet Time Faster, Cheaper, Worse?

June 6, 2003

Main Points

• Quality work is more predictable
• If it does not have to work, anybody can

deliver software on time

• There are known methods – PSPSM, TSPSM

– building motivated and committed teams – changing engineers’ behavior

Personal Software Process, PSP, Team Software Process, and TSP are service marks

• f Carnegie Mellon University.

Outline

• Poor state of software practice
• Need for rational management
• Transforming the way software is done
• The AIS transformation
• AIS TSP project
• The transformation steps

State of the Practice – Projects (1)

• Process-enabling information technology

projects – ERP, SCM, CRM

• Only 33% of outcomes viewed as positive
• Only 58% of all positive outcomes finished

both on time and within budget

Source: “Getting Value From Enterprise Initiatives” Boston Consulting Group, March 2000

State of the Practice – Projects (2)

Ref: CHAOS Study, Standish Group, Summer 2001

Projects late and

• ver budget: 49%

Projects canceled before completion: 23% Projects completed

• n time and on budget: 28%

31% 53%

State of the Practice High Profile Corporate IT Failures

• FoxMeyer
• Hershey Foods Corp.
• Nike
• Rich-Con Steel
• W.W. Grainger
• Whirlpool

State of the Practice- Metaphors

Watts Humphrey profiled in issue of December 6, 1999

• Executives accept schedule commitments when the

engineers offer no evidence that they can meet these commitments.

• Engineers agree to dates when they have no idea how to

meet them.

• Project managers concentrate on the work to be done and

pay little or no attention to the disciplines with which the work is done.

The Consultants’ Checklist (1)

• Begin with a strategic vision and clear objectives
• Do enough up-front work to determine the best options
• Make decisions based on business requirements, not the

capabilities of the software packages

• Divide the initiative into focused, manageable modules

with quick payback

• Choose vendors carefully, and manage them vigorously
• Understand risks and manage them
• Ensure scope does not become too big
• Secure the support of key executives

Source: “Getting Value From Enterprise Initiatives” Boston Consulting Group, March 2000

The Consultants’ Checklist (2)

• User involvement
• Executive management support
• Experienced project manager
• Clear business objectives
• Minimizing scope
• Agile requirements process
• Standard software infrastructure
• Formal methodology
• Reliable estimates

Source: Chaos Success Factors Standish Group, March 1995

The Consultants’ Checklist (3)

• Treat implementation as a business change effort
• Devote the necessary resources to the project
• Make sure that goals, scope and expectations are

clear at the outset

• Track the project’s progress results and scope
• Test the new system every way you can before it

goes live

• Secure top management commitment

Source: “When Too Much IT Knowledge Is a Dangerous Thing” Andrew McAfee, MIT Sloan Management Review, Winter 2003

IT on Internet Time (1)

• Cost of poor project management

– The annual IT Application Development spend is estimated at $250 billion – Annual number of software projects estimated to be around 175,000 – American companies and government agencies will spend $81.2 billion on cancelled projects – In large companies, only 9% of the projects are completed on time and within budget – About 53% of projects will cost 189% of original estimate – Standish Group estimates that lost opportunity costs while not measurable, could easily be in the trillions of dollars

IT on Internet Time (2)

• Cost of poor software quality

– High paid workforce employing 697,000 software engineers and 585,000 programmers – Software defects cost global business an estimated $175 billion in 2001 – NIST study estimates that potential cost reduction from feasible improvements in software quality is $22 billion.

Projected growth in software security flaws

5000 10000 15000 20000 25000 30000 35000 40000 1997 1998 1999 2000 2001 2002 2003 2004 2005 Actual Projected

• Forbes ( 0 6 / 1 0 / 0 2 ) Vol. 1 6 9 , No. 1 3 , P. 1 2 5 ; Goldm an, Lea

IT on Internet Time (3)

IT On Internet Time (4)

• The growth of information technology (IT), added

interconnectedness, and universal access have enabled hackers and would be terrorists to attack critical infrastructures worldwide

• “Wireless technology has the potential of enabling

malicious code to jump off our computer networks and into our everyday lives in a way it never has before” – Antivirus Engineer

Source: IEEE Computer Cover Feature December 2000

IT On Internet Time (5)

Layer Behavior Security

Physical Disruption Antivirus Transport Interception Encryption Network Illegal access Firewall Application Perversion Manual patching

The Opportunity With Software

• There is a better way to manage software
• The better way is also more economical and

faster

• Requires top management to personally lead

the organization transformation

• Quality comes first, even before schedule
• Quality software is produced by disciplined

and motivated professionals

Transformation Principles (1)

• The performance of an engineering organization is

determine by the performance of its teams.

• The performance of an engineering team is

determined by the performance of the engineers.

• The performance of the engineers is determined by

their personal practices.

Transformation Principles (2)

• Management must motivate engineers to adapt best personal

practices – establishing and meeting quality plans – making and meeting development commitments

• When management empowers their teams, the engineers can

– direct their own work – meet business and user needs – produce extraordinary products

• PSP and TSP show engineers how to work this way

Source: “What Next”, Watts Humphrey, SEETAC’03

Personal Software Process, PSP, Team Software Process, and TSP are service marks

• f Carnegie Mellon University.

Changing Engineers’ Behavior

PSP0

• Current process
• Basic measures

PSP1

• Size estimating
• Test report

PSP2

• Code reviews
• Design reviews

PSP3

Cyclic development

Team Software Process

• Teambuilding
• Risk management
• Project planning and tracking

PSP2.1

Design templates

PSP1.1

• Task planning
• Schedule planning

PSP0.1

• Coding standard
• Process improvement

proposal

• Size measurement

Building Motivated and Committed Teams

• The TSP team launch is the most important single

step in building motivated and committed teams

• In the TSP launch, teams make plans and

commitments

• Senior executive participates in the TSP launch

– to help the teams understand what the business needs and trust them to think creatively about their work – give the teams aggressive goals and have them make plans to meet these goals

The TSP Launch Process

Day 1

• 1. Establish

product and business goals

• 2. Assign roles

and define team goals

Day 2

• 4. Build top-

down and next-phase plans

• 5. Develop

the quality plan

• 6. Build bottom-

up and balanced plans

Day 3

• 7. Conduct

risk assessment

• 8. Prepare

management briefing and launch report Launch postmortem

Day 4

• 9. Hold

management review New teams: TSP process review

• 3. Produce

development strategy

TSP and Personal Discipline

• The TSP requires that all the engineers on a team

be PSP trained

• The TSP requires that every team member

measure, plan, and track his or her personal work

Faster, Better, Cheaper and TSP(1)

Accelerating the work:

– The engineering team must make detailed and comprehensive plans – Engineers must make the plans not the managers – Engineers know how to plan – Managers must review the plan for detail and completeness

Faster, Better, Cheaper and TSP(2)

Improving quality:

– Most effective quality strategy is to strive for quality before testing begins – The engineer who developed a program is best able to find and fix its defects – Engineers must be trained in quality methods – They must use these methods on the job – If you do not manage software quality, nobody else can

Faster, Better, Cheaper and TSP (3)

Cutting costs by improving productivity:

– Only way is to maximize the time engineers spend actually working on the project, i.e. task- time – Four key aspects of task-time improvement are:

• Task-time measures
• Engineer motivation
• Task-time plan
• Management review and support

AIS Measured Results

• Our average project schedule overrun has been

reduced from 112% to 5%

• Our average budget overrun from 87% to – 4%
• We now average < one defect per 10,000 lines of

code in delivered products

• One half of the products delivered in the past three

years have been defect free.

The AIS Transformation (1)

Schedule Deviation Individual Value Control Chart - Commercial Systems

• 150
• 100
• 50

50 100 150 200 250 300 350

01/88 01/89 01/90 01/91 01/92 01/93 01/94 01/95 01/96 01/97 01/98 01/99 01/00

Date of Project Phase Start % Deviation

Individual Data Points Mean Upper Natural Process Limit Lower Natural Process Limit One Standard Deviation

The AIS Transformation (2)

Effort Deviation Individual Value Control Chart - Commercial Systems

• 200
• 150
• 100
• 50

50 100 150 200 250 300 350 400 450 500

01/88 01/89 01/90 01/91 01/92 01/93 01/94 01/95 01/96 01/97 01/98 01/99 01/00

Date of Project Phase Start % Deviation

Individual Data Points Mean Upper Natural Process Limit Lower Natural Process Limit One Standard Deviation

TSP Project - Earned Value

Cumulative Earned Value

0.0 10.0 20.0 30.0 40.0 50.0 60.0 70.0 80.0 90.0 100.0 4/22/02 5/6/02 5/20/02 6/3/02 6/17/02 7/1/02 7/15/02 7/29/02 8/12/02 8/26/02 9/9/02 9/23/02 Weeks Earned Value Cumulative Planned Value Cumulative EV Cumulative Predicted Earned Value

TSP Project - Size

PLAN

• SRS - 124 pages
• HLD – 232 pages
• CODE

1.2 – 12584 LOC 1.21 - 17484 LOC ACTUAL

• SRS - 121 pages
• HLD – 188 pages
• CODE

1.2 – 24216 LOC 1.21 - 30081 LOC

Note: 1.21 is, 1.2 + Mini Release

TSP Project - Defect Removal

Defects Removed by Phase for PCS 1.21 (PCS 1.2 & Mini Release)

15.3 48.2 208.2 12.8 47.2 29.5 55.0 116.0 173.0 39.0 121.0 82.0 0.0 0.0 2.6 0.0 25.8 185.6 97.0 3.0 0.0 5.0 30.0 173.0 0.0 50.0 100.0 150.0 200.0 250.0 REQ Insp HLD Insp DLD Rev DLD Insp Code Code Rev Compile Code Insp U T IT ST AT Phase Defects Removed by Phase Plan Actual

TSP Project - Process Quality

59.46 43.24 51.35 91.89 91.89 100.00

0.00 25.00 50.00 75.00 100.00

Percentage of Modules

COMPILE UNIT TEST INTEGRATION TEST SYSTEM TEST ACCEPTANCE TEST POST DELIVERY

Phases PCS - Percent Defect Free

TSP Project - Productivity

44.25 33.04 36.53 32.19 34.79 40.68 27.67 33.33 35.39 41.96 36.94 33.87 39.41 37.94 36.73

0.00 10.00 20.00 30.00 40.00 50.00 Priya Anu Bupesh Raja Bala Jeasthri Sangeetha Thenmani Rama Asokan Philip Suganthi Ezhil Anil Sunita

• Avg. Productive Hours per Week

Average Schedule Deviation - Range

• 20%

0% 20% 40% 60% 80% 100% 120% 140% 160% Pre TSP/PSP With TSP/PSP

Average Effort Deviation - Range

• 20%

0% 20% 40% 60% 80% 100% 120% Pre TSP/PSP With TSP/PSP

Defects/KLOC in Acceptance Test - Range

0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 Pre TSP/PSP With TSP/PSP

System Test Duration (Days / KLOC) - Range

1 2 3 4 5 6 7 Pre TSP/PSP With TSP/PSP

http://www.sei.cmu.edu/publications/documents/00.reports/00tr015.html

PSP/TSP - Quiet Quality Revolution (1)

PSP/TSP - Quiet Quality Revolution (2)

• “You actually get your money back after 1200

lines of code”

• “The data says that if you follow this, you get

absolutely the cheapest time to market”

• “The first company to really get this right will

have a competitive advantage nobody will be able to keep up with”

• “It is so revolutionary that I remember the exact

date I was introduced to it”

• “It is hard to believe unless you do it”
• Source: IEEE Software, November/December 2000

The Transformation Steps

• Establish a quality policy
• Kick-start SPI
• Set the long term direction – PSP,TSP
• Motivate managers and engineers
• Ensure changes are maintained
• Ensure people use the new methods in their

work

Contact Information

Girish V. Seshagiri CEO Advanced Information Services Inc. Co-founder The Watts Humphrey Software Quality Institute Websites: www.advinfo.net www.watts-sqi.org E-mail: girish@advinfo.net