lean software management: bbc worldwide case study submitted to professor shervin shirmohammadi in...
TRANSCRIPT
Lean Software Management:BBC Worldwide Case Study
submitted to Professor Shervin Shirmohammadi in partial fulfillment of the requirements for the course ELG 5100
submitted byTaiyan Yu (7287759) & Menglan Liu (6836472)
Outline
• Introduction of Lean thinking• Research Methodology• Reliability of The Data Collected• Digital Hub Team• Performance Data and Analysis• Additional Tips• Future Works• Conclusion
Lean ThinkingSeven principles[3] [4]:• Eliminate waste• Amplify learning• Decide as late as possible• Deliver as fast as possible• Empower the team• Build integrity in• See the whole
Toyota Production System(TPS) [1]
SpaceManpowerMaterialstime
Research Methodology
• Hypothesis Implementing lean practices Reduce: lead times error rates variability• Method Observation + Phone calls + E-mails
Research Methodology(…)
• Data were collected from: 1) Digital Hub (Digi-Hub) Team 2) Semi-structured interviews 3) Kanban boards 4) Recording the operation of lean system 5) Daily “stand-up” meeting 6) Review the outputs
Reliability of The Data Collected
• Time Line Implementation: April 2008 Data: August 2008 (collected) October 2008 (used)
Reliability of The Data Collected(…)
• Size and Volume of Work Started Minimum Marketable Features (MMFs)
Reliability of The Data Collected(…)
• Complexity of Work• Governance Arrangements• Composition of Team• Engineering Practices
Digital Hub Team• Office Layout and Work Flow: two kanban board
s (A,B) and four information radiators (C,D,E,F)
Digital Hub Team(…)
Digital Hub Team(…)
• Daily Standup
Digital Hub Team(…)
Release notification and daily support process tasks
Record decisions
Vote
Performance Data and Analysis
• An integral part of any lean manufacturing implementation is a variability reduction effort, to enable a process to achieve the same (or greater) throughput with less WIP
• Throughput = WIP/cycletime
average output
the inventory between the start and finish points of a production process
the time a unit spends as WIP
Performance Data and Analysis1. Lead Time: • the total elapsed time from when a customer requests software to when t
he finished software is released to the customer.• kanban board A: “decomposed engineering ready” to kanban board B: “re
lease ready”.
results show software is being delivered with 47% less variance and on average 37% quicker.
Performance Data and Analysis
2. Development Time: is recorded in working days, from kanban board B stages: Dev. Ready to Dev. Complete.
The variance of development time fell by 78% from 30.5 days to 6.8 days. The mean development time was reduced 73% from 9.2 to 2.5 working days over the nine months.
3. Release Frequency Per Month: the number of items released to customers.
Performance Data and Analysis
a blanket release freeze to ensure complete priority was given to the production of year-end financial data
4. Live Defects Per Week: • the bugs reported by customers during a week plus the bugs still open.• recorded on red kanban cards and added to the Dev. (Development) stage
of kanban board B.
Performance Data and Analysis
Variance fell by 33% as the upper control limit reduced from 7.6 to 5.1 open bugs per week.
Defects still open and reported each week fell by 24% from 2.9 to 2.2
Performance Data and Analysis5. Continuous Improvement : issues identified and time to r
esolve.The outlier in 2008 was a result of waiting for a third party to complete their work (a s
pecial cause).
Over the 12 months, the mean number of working days itemswere blocked was reduced by 81% from a mean of 25.8 days to 4.9 days per month.
Additional Tips
• 1. A master kanban board can be used to record and summarize the progress of all the smaller projects.
• 2. Lean requires a stable, experienced team with low staff turnover and a project manager who knows the skills and abilities of their team.
• 3. Lean is not a substitute for professional software engineering practice.
Future Works
• The time taken to move on kanban board A from “proposed ideas” to “decomposed engineering ready” is currently not recorded, and this time could be significant.
• Lead time could potentially be further reduced by make UAT easier for customers.
Conclusion
• The volume of work that was allowed to enter the process
• The reduced cycle time• Continuous improvement carried out by the te
am on a daily basis• Statistical process control
Reference
[1] J. M. Morgan and J. K. Liker. The Toyota Product Development System: Integrating People, Process and Technology. New York: Productivity Press,2006.[2] Er. Kirtesh. Jailia, Mrs. Sujata, Mrs. Manisha. Jailia, Mrs. Manisha. Agarwal, Lean Software Development(“As a Survival Tool in Recession”), International Journal of Software EngineeringandItsApplicationsVol.5No.3,July,2011[3] M. Poppendieck and T. Poppendieck, Lean Software Development: An Agile Toolkit. Boston, MA:Addison-Wesley,2003.[4] Mary Poppendieck, MichaelA. Cusumano, Lean Software Development: A Tutorial, IEEE Software,Sep/Oct,2012.[5] Scrum-ban: http://leansoftwareengineering.com/ksse/scrum-ban/
Thank you:)