Intro to Kanban - AgileDayChile2011 Keynote
•
42 likes•9,060 views
Keynote from Dabid J. Anderson (@agilemanager) on AgileDayChile 2011, november 17 2011
Recommended
More Related Content
What's hot
What's hot (20)
Viewers also liked
Similar to Intro to Kanban - AgileDayChile2011 Keynote
Similar to Intro to Kanban - AgileDayChile2011 Keynote (20)
More from ChileAgil
More from ChileAgil (20)
Recently uploaded
Recently uploaded (20)
Intro to Kanban - AgileDayChile2011 Keynote
- 1. An Introduction to Kanban What is it & why would I use it? David J. Anderson David J. Anderson & Associates dja@djandersonassociates.com
- 2. Book Published April 2010 A 72,000 word intro to the topic
- 3. German published January, 2011 Translation by Arne Roock & Henning Wolf of IT-Agile
- 4. Spanish published May 2011 Translated by Masa K Maeda, PhD
- 5. Portuguese (Brazilian) published October 2011 Translation by Andrea Pinto Recife, Brazil
- 6. http://leankanbanuniversity.com http://www.limitedwipsociety.org LinkedIn Groups: Software Kanban Yahoo! Groups: kanbandev Yahoo! Groups: kanbanops
- 8. Taiichi Ohno… TPS = JIT + Autonomation Kanban is the tool that enables these
- 9. What is a kanban (pull) system?
- 10. An example of a virtual kanban system overlaid on a software development process PTC Eng Mgr Change PM Requests Developers Testers Kanban Kanban 8 cards 8 cards (3 WIP 5 Queue) User Acceptance Test Product Backlog 25 Days Managers
- 11. We started visualizing these flows on white boards and holding standup meetings in front of it Provides the 1st (team) level of process capability feedback
- 12. WIP limits create a pull system and white board provides visualization of flow WIP Limit – regulates “inventory” at each stage in the process Pull Flow – from Engineering Ready to Release Ready
- 13. Queue Replenishment & delivery run on independently determined cadence Queue Replenishment Delivery Cadence How often can we How often can reasonably meet with customers (or business downstream functions) stakeholders? economically take delivery?
- 14. Colors represent classes of service Expedite Fixed Delivery Date Significant delay incurred on or from a specific date in near future Standard Class (Near) linear cost of delay beginning immediately Intangible Class No tangible cost of delay within reasonable lead time to delivery window
- 15. We discovered that standup meetings could be scaled to a large size In this example more than 40 people attend a standup for a large project with 5 concurrent development teams. The meeting is usually completed in approximately 10 minutes. Never more than 15.
- 16. Major Project with two-tiered kanban board using swim lanes for each development team Swim lanes control WIP limit – Requirement WIP (green) controlled by number of lanes
- 17. Monthly Operations Review using quantitative measures of capability & demand provides the 2nd (organization) level of process feedback
- 18. “After Meetings” after stand-ups to focus on immediate process issues
- 19. Why do we pursue process improvements at all?
- 20. Goals for improvement initiatives Economically balance capability against demand
- 21. Available options Process
- 22. Most process geeks & IT managers are operating over here Process
- 23. Agile methods have been the favorite approach to improving capability in the 21st Century!
- 24. However, some people & organizations have resisted adoption of Agile methods!
- 25. If not every organization is ready to adopt an Agile method, how can we encourage them to become more Lean?
- 26. Kanban is the gateway to a Lean organization
- 27. So how do we go about introducing Lean into organizations that have failed to adopt an Agile method such as TDD?
- 28. The counter-intuitive answer is to use a pull system that limits work-in-progress to catalyze the introduction of other Lean concepts
- 29. So What is the Kanban Method?
- 30. My motivation for adopting kanban systems was to prevent overburdening, & control variability that affects flow and encourage an evolutionary approach to change
- 31. In developing the Kanban Method, a change management approach that uses kanban systems to provoke change, we are enabling the emergence of Lean software development in organizations
- 32. How does the Kanban Method work?
- 33. Kanban is based on 3 principles 1. Start with what you do now 2. Agree to pursue incremental, evolutionary change 3. Initially, respect current roles, responsibilities & job titles
- 34. Then… then adopt 5 core practices that are observed to be present in successful Kanban implementations
- 35. 5 core practices for successful Kanban adoption Shallow 1. Visualize 2. Limit Work-in-Progress Depth 3. Manage Flow 4. Make Process Policies Explicit 5. Improve Collaboratively (using models & scientific method) Deep
- 36. It‟s not a question of right or wrong … Shallow It‟s a question of shallow or deep! Depth Shallow implementations tend to produce fewer, less dramatic results Deep
- 37. When… When all 5 core practices are adopted they form the seed conditions for complex Kanban as a adaptive system that enables a Lean(er) way of working to emerge
- 40. Limit Work-in-Progress 20 3 2 4
- 41. Observe Flow (empty test column)
- 42. Observe Flow with a CFD Device Management Ike II Cumulative Flow 240 220 200 180 160 Features 140 120 Avg. Lead Time 100 80 60 WIP Avg. Throughput 40 20 0 ar ar eb eb eb ar ar ar M M -M -M -M -F -F -F 2- 9- 10 17 24 16 23 30 Time Inventory Started Designed Coded Complete
- 43. From the simple geometry we can observe Little‟s Law WIP Throughput = Lead Time From observed capability
- 44. Observe Lead Time with a control chart
- 45. Observe Flow with a spectral analysis histogram of lead time Lead Time Distribution 3.5 3 2.5 CRs & Bugs 2 1.5 1 0.5 0 6 3 0 7 4 1 8 1 8 15 22 29 36 43 50 57 64 71 78 85 92 99 10 11 12 12 13 14 14 Days SLA expectation of Mean of 51 days with 98% on-time 31 days SLA expectation of 44 days with 85% on-time
- 46. Development is a Bottleneck This is an example of using a model to identify an improvement opportunity
- 47. Analysis is overloaded Analysis suffers from non-instant availability of subject matter experts / business owners
- 48. Couple observation of non- instant availability of expertise with visual & quantitative evidence of muri in flow to encourage better availability
- 50. Leadership is the magic ingredient sprinkle liberally over the 5 seed properties
- 51. The WIP limit provokes the conversation
- 52. Without a WIP limit the Idle & Stuck comments may never emerge
- 53. The team has a choice to break the WIP limit and ignore the issues, or face up to the issues and address them using the models
- 54. The WIP limit simply provokes the conversation. Leadership encourages discussion about improvement. Use of Models and other evidence leads to an improvement suggestion and implementation
- 55. Big Projects with Kanban
- 56. Major Project with two-tiered kanban board using swim lanes for feature sets Swim lanes control WIP limit – Requirement WIP (green) controlled by number of lanes
- 57. Kanban daily standup meetings can be very large In this example more than 40 people attend a standup for a large project with 5 concurrent development teams. The meeting is usually completed in under 15 minutes
- 58. Spontaneous Quality Circles form after the standup to focus on immediate process issues Daily standup provides forum for spontaneous association to attack process issues affecting productivity and lead time 3 day freeze on test environ was a transaction cost on release that caused a bottleneck at “build” state. Reduced to 24 hrs. Result was improved smoother flow resulting in higher throughput and shorter lead time.
- 59. Sticky Buddy scheme was instituted to allow remote workers to keep kanban board up-to-date and synchronized with electronic tracking “Cancelled” area With trash can For partially worked Items obe.
- 60. Estimating and planning major projects
- 61. Cumulative Flow and Predictive Modeling with S-Curve Device Management Ike II Cumulative Flow 240 220 200 180 160 Features 140 Typical S-curve 120 100 80 60 40 20 0 ar ar eb eb eb ar ar ar M M -M -M -M -F -F -F 2- 9- 10 17 24 16 23 30 Time Inventory Started Designed Coded Complete
- 62. Simulating S-Curve with a Z Device Management Ike II Cumulative Flow 240 220 60% 200 180 160 Slope in middle Features 140 3.5x - 5x slope 120 100 at ends 5x 80 60 20% 20% 40 20 0 ar ar eb eb eb ar ar ar M M -M -M -M -F -F -F 2- 9- 10 17 24 16 23 30 Time Inventory Started Designed Coded Complete
- 63. Track actual throughput against projection Device Management Ike II Cumulative Flow 240 220 200 180 160 Features 140 120 100 80 Track delta between 60 40 planned and actual 20 each day 0 ar ar eb eb eb ar ar ar M M -M -M -M -F -F -F 2- 9- 10 17 24 16 23 30 Time Inventory Started Designed Coded Complete
- 64. Variability in Throughtput (velocity) It is important to understand the role throughput plays in long term planning in Kanban but why it is not useful for short-term goal setting Often velocity exhibits a +/-2x spread of variation As a result velocity cannot be used as a short- term planning tool See following examples
- 65. Velocity Variation South African Team from 2011plotted per Sprint (2 weekly) Mean 29, UCL (+1 sigma) 43 (+1.5x), LCL (-1 sigma) 15 (- 2x)
- 66. DBA Team Velocity 90 80 70 Trend 60 50 Total Velocity Small support tasks 40 30 20 Trend (not included in total velocity) 10 Week of Christmas 0 Mattias Skarin client based in Paris in 2009/2010, plotted weekly Mean 42, +1 sigma = 55, -1 sigma = 29 (+/- 1.4x)
- 67. Investment Bank, London, Extreme Programming Weekly Mean 10, Max = 16, Min = 6 Spread (+/- 1.6x)
- 68. Velocity Control Charts Completion Velocity Chart 40 30 UCL 29.2 20 Completion Completion Velocity Velocity UCL 10 CL 7.206896552 Sigma +2 +1 Sigma 0 -10 LCL -14.8 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 -20 Date Motorola PCS 2003, FDD project plotted Daily Mean 7.2, +1 sigma = 15, -1 sigma = 0 (+/- 2x) Weekly Mean 36, +1 sigma = 63, -1 sigma = 11 (+ 1.7x, -3.3x)
- 69. Variability in scope/requirements It is also important to realize how much variability there is in the scope/scale or requirements and how this must be accommodated in our plans
- 70. Unplanned Work Report Scope Creep Dark Matter (emergent features) Original Scope Dark matter planned as a 19% expansion over original scope Actual Dark Matter over final original scope is 26% Total scope compared to original commitment is 13% greater
- 71. Typical Agile teams using User Stories analysis produce 40-60% dark matter Stories are recognized as Epics and broken into more stories. The customer does not consider the scope to have changed
- 72. TV/Movie Company in USA 2008 Initial Scope is 125 story points Within days this total scope reaches 190 due to dark matter expansion Management intervened on 4/21 to stop dark matter (deferring future scope to product backlog) Observed dark matter expansion is 52% but real number was much greater
- 73. Original CFD shows same top line Dark matter quotient is 19% Device Management Ike II Cumulative Flow 240 220 200 180 160 Features 140 120 100 80 Track delta between 60 40 planned and actual 20 each day 0 ar ar eb eb eb ar ar ar M M -M -M -M -F -F -F 2- 9- 10 17 24 16 23 30 Time Inventory Started Designed Coded Complete
- 74. Make a long term plan to build platform replacement Device Management Ike II Cumulative Flow Required throughput (velocity) 240 220 200 180 160 Features 140 Slope in middle 120 3.5x - 5x slope 100 at ends 5x 80 60 40 20 0 2006 2008 ar ar eb eb eb ar ar ar M M -M -M -M -F -F -F 2- 9- 10 17 24 16 23 30 During the middle 60% of the project schedule Time we need Throughput (velocity) to average 13 Inventory Started Designed Coded Complete features per month
- 75. Little‟s Law Determines staffing level Target to achieve plan WIP Throughput = Lead Time From observed capability Treat as Fixed variable
- 76. Changing the WIP limit without maintaining the staffing level ratio represents a change to the way of working. It is a change to the system design. And will produce a change in the observed „common cause‟ capability of the system
- 77. Plan based on currently observed capability and current working practices. Do not assume process improvements. If changing WIP to reduce undesirable effects (e.g. multitasking), get new sample data (perform a spike) to observe the new capability
- 78. Little‟s Law Determines staffing level Target to achieve plan WIP 13 / month = 0.25 months WIP = 3.25, round up to 4. Might be safe to From observed capability round down to 3. If current working practice is 1 unit WIP per person then 3 people are needed to work this project exclusively
- 80. Educate the workforce to recognize process problems that affect performance
- 81. Manage bottlenecks to increase throughput Known max productivity Idea Analysis Design Code ~90 ~80 ~50 ~50 Error Error Error Working Acceptance System Unit Code Test Test How do I trim 25% Test ~80 ~30 ~40 ~50 off the schedule for Quality a project going Schedule 1. Identify - Current CCR is System Test – 30 through this system? per month 2. Exploit - Testers relieved of all non-essential tasks, extra PMs assigned to complete administrative tasks, analysts assigned to future test plans 3. Subordinate - Requirements release restricted to ~30 per month Quality 4. Elevate - Plan to recruit temporary testing staff immediately Schedule
- 82. Non-instant Availability Looks like a bottleneck Same thinking process applies Management approach is similar but policies will be different depending on type of bottleneck CCR vs NIA
- 83. Wimbledon comes in on time every year How do event planners do it? Scheduling Wimbledon isn’t an exact science Except 2004 when it rained on Sunday and the final was on Monday ;-) Games last different lengths of time and weather conditions can stop play altogether but the Men’s Final always happens on the 2nd Sunday If only we could project manage like this!
- 84. Quantity of pink issue tickets on board directly indicates flow impacting problems that need attention from management Issues are the exception – attached to work items that are blocked for external reasons and call attention to problems preventing smooth flow
- 85. Focusing on efficiency produces better cost accounting results for large batch sizes
- 86. But there are also coordination costs in knowledge work problems Communication Queue Task W Queue Task Wait Setup Cleanup Queue Task Wait Q Task Wait Time
- 87. And in knowledge work problems, coordination costs grow non-linearly with batch size Communication Queue Task W Queue Task Wait Setup Cleanup Queue Task Wait Q Task Wait Time
- 88. Sources of Economic Costs Transaction Costs Coordination Costs Failure Demand
- 89. Transaction Coordination Costs Transaction cost Costs Costs Value-added Work Failure Demand time
- 90. Good Bad
- 91. Conclusions
- 92. Limiting work-in-progress can catalyze incremental changes
- 93. respect The team must the WIP limit and value the conversations it provokes about problems
- 94. Leadership is the secret sauce! Encourage it from any team member regardless of position, experience or authority
- 95. Enable the team with transparency of process (visualization of invisible work & process dynamics.) Use models for understanding problems and improvements will occur.
- 96. These improvements will provide better economic and sociological outcomes
- 97. Enabling our goal to… Economically balance capability against demand
- 99. Thank you! dja@agilemanagement.net http://www.agilemanagement.net/
- 100. About… David Anderson is a thought leader in managing effective software teams. He leads a consulting firm dedicated to improving economic performance of knowledge worker businesses – improving agility, reducing cycle times, improving productivity and efficiency in technology development. He has 25+ years experience in the software industry starting with computer games in the early 1980‟s. He has led software teams delivering superior productivity and quality using innovative agile methods. He developed MSF for CMMI Process Improvement for Microsoft. He is a co-author of the SEI Technical Note, CMMI and Agile: Why not embrace both! David‟s book, Agile Management for Software Engineering – Applying the Theory of Constraints for Business Results, introduced many ideas from Lean and Theory of Constraints into software engineering. David was a founder of the Lean Software & Systems Consortium, a not for profit dedicated to promoting better standards of professionalism and effectiveness in software engineering. Email… dja@agilemanagement.net
Editor's Notes
- Kanban board gives visibility into process issues – ragged flow, transaction costs of releases.Daily standup provides forum for spontaneous association to attack process issues affecting productivity and lead timeFor ex: 3 day freeze on test environment was a transaction cost on release that caused a bottleneck at “build” state. Reduced to 24 hrs. Result was improved smoother flow resulting in higher throughput and shorter lead time.