Problems and Software Process Improvement

1
Problems and Software Process Improvement

• Jim Kelly
• Zenkara Pty Ltd
• jim_at_zenkara.com

2
Outline

• Present an overview of the research
• Describe survey conducted
• Describe results and feedback
• Provide some analysis and
• Generate some discussion

3
Overview of the research 1

• What were looking at, and what were not.
• In our scope we have reviewed the problems that
  CMMI (or an alternative SPI model) addresses
• We have not included the actual problems
  experienced in deploying SPI initiatives (though
  this is related)

Where CMMI is referred to, you can generally
substitute alternate SPI models/methods
4
Overview of the research 2

• Examining the problems addressed by using
  Software Process Improvement models/methods
• Hypothesis is that some organisations do not
  identify the specific goals that CMMI is intended
  to address
• If organisations better defined the specific
  goals that CMMI is to address, there would be
  more successful deployments

5
Overview of the research 3

• Prime focus is on SW CMM and CMMI for software
  organisations (although other models are being
  examined)

http//www.software.org/quagmire/
6
Details of survey 1

• Conducted interviews with 12 Software Process
  Improvement experts who operate in Australia
  either embedded in software organisations or
  externally who provide advice to software
  organisations.
• Considered numerous models/methods
• SW CMM, CMMI, ISO9001, SPICE
• Results were qualitative and descriptive

7
Details of survey 2

• Conducted face to face and phone interviews (45m
  to 1.5hrs) with an indicative line of questions
• How long have you been active in CMMI?
• What training have you had in CMMI?
• What is your rationale for using the CMMI?
• What benefits of the CMMI were identified to
  contribute to your organisation?
• What tools were used to identify the problems?
• How do you know they were appropriate?
• CMMI was used for the majority of interviews,
  but was broadened for some organisations to SPI.

8
Results and Feedback 1

• For large organisations (gt100 staff), PI model is
  mandated by corporate policy or by customers.
• Benefits of the CMM/CMMI and other models are
  often not identified
• Costs of PI model deployment often not exhaustive
  - e.g. only recording PI staff effort, and not
  operational involvement

9
Results and Feedback 2

• Organisations which use SW CMM or CMMI generally
  also use at least one other SPI model/method
  (e.g. ISO9001, EFQM, SPICE, Personal Software
  Process, Team Software Process)
• Records of initial justification for initiating
  Process Improvement initiatives often not kept

10
Results and Feedback 3

• Some organisations have links from their
  Process Improvement efforts to their
  organisations Business Goals/Objectives
• However the links are mostly tenuous with little
  details about what the PI effort will do to e.g.
  improve productivity, schedule, budget, product
  quality.

11
Results and Feedback 4

• No real structured approach to finding and
  understanding problems to be solved.
• Most experts interviewed had an unstructured view
  of organisational problems very much a mess
• Ackoffs definition of mess world.std.com/lo/9
  6.02/0009.html

12
Results and Feedback 5

• Many experts not immediately aware of top 5
  problems with their organisation
• Many also did not have knowledge of specific
  problems which have been solved over the past 1-2
  years

13
Results and Feedback 6

• Institutionalisation is often seen as most
  important problem (particularly in mature
  organisations!?)
• CMMI experts who are in an immature organisation
  are aware of the problems and are aware that the
  problems are not being addressed in a structured
  approach, but do not use formal techniques
  (Pareto, SPC).

14
Initial analysis 1

• Rationale for selecting specific SPI models is
  far from clear in many organisations.
• Links with business goals must be made
  explicit how exactly will the PI contribute?
• Simply saying improve productivity or reduce
  rework is not adequate e.g. what component or
  process of the CMMI precisely is going to help?

15
Initial analysis 2

• clearly identify benefits
• How they were calculated
• Which specific CMMI actions helped and how
• Use the impact of the CMMI to help identify
  current problems and areas for improvement

16
Initial analysis 3

• Look at the current effort in multi-SPI-model
  initiatives
• Interactions
• Efficiencies
• Quite often improving something which satisfies
  requirements of one model/method will also
  satisfy part of another.

17
Initial analysis 4

• More mature organisations (CMMI L3) often only
  see problems within the scope of the CMMI

18
Useful Websites

• www.sei.cmu.edu
• www2.umassd.edu/SWPI/1docs/SPResearch.html
• www.sqi.gu.edu.au
• www.psmsc.com
• www.swquality.com/users/pustaver/index.shtml
• nicta.com.au/uploads/documents/Why20Organisations
  _20Dont_20Use20CMMI.pdf
• www.dtic.mil/ndia/2004cmmi/CMMIT7Thur/CMMIUsersGro
  upjmsiviy.pdf
• www.stsc.hill.af.mil/crosstalk/2002/02/menezes.htm
  l

19
Demonstration

• Break into groups of two preferably with
  someone you dont know
• Introduce yourself
• Individually read and answer the red sheet
• Review your team members problems
• Try to come to a decision about which problem
  your team would identify as 1 priority

20
QA

• Does this correlate with your experiences?
• What is missing?
• What do you interpret differently?
• Which models do YOU use?

21
Who are we?
Zenkara provides advanced technology for
mission-critical software engineering companies.
Head Office 2/27 Waverley Street Annerley Qld
4064 Australia PO Box 1622 Milton QLD
4064 Mobile[phone] direct 61 (0) 7
3892 6771 email jim_at_zenkara.com web
http//www.zenkara.com ABN 85 050 409 660