Lssc12.emergent.patterns.in.it
3 likes1,070 views
The document discusses emergent patterns for using Kanban systems in IT operations. It addresses common problems IT operations teams face related to dependencies, specializations within teams, and interruptions. It provides examples of how to make these issues more visible using Kanban techniques, such as tracking dependency risks, illustrating different specialization levels, and monitoring interrupt sizes. The overall point is that increasing visibility of these challenges helps improve risk recognition and better understand and meet demands on IT teams.
Lssc12.emergent.patterns.in.it
- 1. Emergent Patterns for Kanban Systems in IT Operations Dominica DeGrandis @dominicad dominica@djaa.com Lean Systems & Software Boston, MA May 2012
- 2. Emergent Patterns for Kanban Systems in IT Operations Dominica DeGrandis @dominicad dominica@djaa.com www.djaa.com kanbanops yahoo group Kanban Roundup
- 3. Things that Plague IT Ops teams 1. How do we not get killed by dependencies? 2. How do we deal with Brent? 3. How can we get important work done with so many interruptions?
- 4. Things that Plague IT Ops teams Next (3) Study (1) Validate Specialization Dependencies Interrupts
- 5. Ge3ng Visibility on Dependencies
- 6. Issue a4ached to change request –
- 7. Electronic Board Dependency tags
- 8. Dependencies carry some level of risk. How much Risk?
- 9. Levels of risk? Outside of control: Team Department Company 3rd party vendor in a different time zone
- 10. Understanding the risk cost? history? criteria?
- 11. Example 1 A dependency on the data team down the hall. They have changes that impact our CI scripts. If we don’t get them by next week, we’ll need a manual step in the build process to upgrade the DB. Strategy?
- 12. Courtesy David Lutz Duplicate ticket from other teams board placed on Ops board.
- 13. Data Team Kanban Board Input Design Data Test Deliver work item tagged as impacting Ops team.
- 14. If an issue comes up, the card gets moved to “blocked” area for extra visibility.
- 15. Electronic Board Dependency tags
- 16. Example 2 A dependency on the marketing department located in a different city. They want new images deployed to the website, but image rights have not been cleared yet. Marketing campaign is tied to next months press release. Strategy?
- 17. Waiting area for items with external dependencies 5 4 4 2 Input Investigate Action Test Complete Queue Doing Done Doing Done Dependency risk placed in separate Waiting for (3) “Waiting For” area for management to deal with.
- 18. WaiHng area for external dependencies 5 4 3 4 2 2 = 20 total Analysis Input Ready Implementation Build Release ... Queue In Prog Done In Prog Done Ready Test Ready Waiting on External Group Dots denote clock ticking on SLA Late against SLA
- 19. Courtesy Ian Carroll Blocked” area for transparency.
- 20. External Blocker Box medium-term blocked Cards at the top are higher priority that cards at the bottom. long-term Blocked Courtesy Ian Carroll
- 21. Design for Visibility on SLA status Time remaining to meet SLA Input Done I I I I I I I I I I I I I Priority
- 22. Fixed date
- 23. Example 3 The CFO promises the owner of the company that we can deliver a regulatory requirement by the end of the year and avoid legal penalties. The request is dependent on input from Company C who has a terrible track record. Strategy?
- 24. Short term tactical plan: Policy set for company C where dependencies are pulled into the ready queue after confirmation of completion on their part.
- 25. Long term strategic fix: Contracts modified with vendors to contain sufficient remedies to ensure SLAs are met.
- 26. Consensus Dependencies need visibility on risk. Show the risk
- 27. Ways to Illustrate Risks for Dependencies 1. Show cost associated with delay. 2. Show projected delay (history/simulation). 3. Display explicit policies for dependencies. 4. Put risk info on work item
- 28. Things that Plague IT Ops teams 1. How do we not get killed by dependencies? 2. How do we deal with Brent? 3. How can we get important work done with so many interruptions?
- 29. Things that Plague IT Ops teams Next (3) Study (1) Validate Interrupts Specialization Dependencies
- 33. PaJern or anH-‐paJern? “… they are willing to take responsibility for the work their own work, but not of the whole team.”
- 34. PaJern or anH-‐paJern? “Upper management wants visibility on individuals work.”
- 35. Kanban Design for Visibility on Specializations Operations Board Investigate Committed Test Delivered SysAdmin Jason Config/Automate Kris Network/Monitor Pat DBA Mark Security Joe Visibility on types of demand
- 36. Highly specialized workforces are not very flexible, but it might be the workforce that enables you to win.
- 37. Skill Level Rankings Example Master Journeyman Practitioner Novice Newb
- 38. Skill Level Rankings Example 10 yrs exp or 10,000 hours of practice, mentors others. Master Knows when to break the rules. Journeyman 6-9 yrs exp, can handle > 80% of requests, trains others. Practitioner 3-5 yrs exp, can handle > 50% of requests. Novice 1-2 yrs exp, in training Newb Newbie
- 39. Tools & skill level of team Puppet Apache Ruby MySQL Oracle Linux
- 40. Things that Plague IT Ops teams 1. How do we not get killed by dependencies? 2. How do we deal with Brent? 3. How can we get important work done with so many interruptions?
- 41. Things that Plague IT Ops teams Next (3) Study (1) Validate Interrupts Dependencies Specialization
- 42. Making Interrupts visible Operations Investigate Implement Buyoff Done Planned 25% Unplanned 75%
- 43. Pattern - The Goalie Handles small interrupts Rotates weekly Expands knowledge base Gains flexibility in team
- 44. Help Desk Board Fire Lane Input Implement Test Delivered Priority 1. 2. 3. 4. 5.
- 45. What is the threshold for creating a ticket? Transaction cost vs. value
- 46. PaJerns for Task Size Policies Example: If < n min, just do it ExcepHons…. 1. Dependency 2. Only Brent knows how to do it 3. Visibility of small tasks in reporHng
- 47. Showing impact of Interrupts Interrupts Investigate Doing Buy off Done Small 6 Medium 3 Large 2 S = 1 day, M = 2-3 days, L= 1+week (LT)
- 48. Caution! Consider the context It’s likely different. Using another design for your situation can be dangerous.
- 49. Things that Plague IT Ops teams Next (3) Study (1) Validate What’s the Dependencies point? Specialization Interrupts
- 50. Things that Plague IT Ops teams Next (3) Study (1) Validate Dependencies Questions? What’s the point? Specialization Interrupts
- 51. What’s the Point? 1. Show the risk Improving recognition of risks in IT is powerful!
- 52. What’s the Point? 1. Show the risk Improving recognition of risks in IT is powerful! 2. Show the types of work Understanding demand helps to meet it.
- 53. What’s the Point? 1. Show the risk! Improving recognition of risks in IT is powerful. 2. Show the types of work Studying/understanding demand helps to meet it. 3. Track interrupts for visibility Making invisible work visible reduces unknowns.
- 54. Things that Plague IT Ops teams Next (3) Study (1) Validate Dependencies Questions? Specialization Interrupts What’s the point?
- 55. Emergent Patterns for Kanban Systems in IT Operations Delivered Dominica DeGrandis Dependencies @dominicad Specialization dominica@djaa.com Interrupts What’s the point? Questions?