Download as PDF, PPTX
Uploaded byMichele Orselli
Continuous, continuous, continuous
The document discusses the principles and practices of Continuous Delivery (CD) in software development, emphasizing its importance in delivering reliable and deployable software through an incremental and iterative approach. It identifies common challenges such as long release times and poor communication, while advocating for cross-functional teams and automated testing to enhance efficiency. The authors propose a maturity model for assessing CD implementation and stress the need for collaboration among development and operations teams.
Continuous, continuous, continuous
- 1. Continuous, continuous, continuous. A better way to delivery better code Michele Orselli - @_orso_ Francesco Trucchia - @cphp
- 2. Francesco Trucchia VicePresidente @ideato @cphp Michele Orselli CTO @ideato @_orso_
- 3. Agenda 1. Motivations 2. Bad smells 3. What is CD 4. Implementing CD
- 4. Delivering working software matters to business people
- 5. 1. Find theright target market 2. Testing the business model 3. Continuous improvement 4. Scaling fast
- 6. Find the right target market
- 7. Testing the businessmodel
- 8.
- 9.
- 10. Bad smells ofsoftware delivery https://www.flickr.com/photos/gotosira/4699308007
- 11. 1. Business knowledgeis not shared 2. Long release time 3. No reality check 4. Long bugs backlog 5. Dev and ops silos
- 12. Business knowledge isnot shared
- 13.
- 14.
- 15.
- 16. Dev and opssilos
- 17. How can youfix these problems?
- 18. Continuous Delivery to the rescue!
- 19. software is alwaysin a deployable state deployable software over adding new features everyone can have feedback on a change one click deploy
- 20. Why CD it’suseful?
- 21.
- 22. Requirements Development QARelease
- 23. where do westart? when are we finished? how do we proceed? !
- 24.
- 25. framework to understandwhere we are two dimensions: process vs app !
- 26. Process optimized ! measured ! consistent ! repeatable ! non repeatable
- 27. Application code |configuration | host | data | test
- 28.
- 29. How to proceed? incremental approach remember: everyone in the team is responsible for software delivery !
- 30.
- 31. no big releases prioritize by business value deliver value on every iteration
- 32. short feedback cycle small releases are easier to manage
- 33. ! but… Iwant all the features!
- 35. short feedback means test your business early short deploys less bugs
- 36.
- 37. changes are integratedearly reduced risks early feedback “if it hurts do it more frequently”
- 38. keep everything inversion control code, doc, data
- 39. you have testsfor you app unit acceptance capacity
- 40. but… tests area waste of time! what about testers?
- 42. automated tests mean higher confidence on your software short deploys less bugs
- 43. CI aims toautomate build process
- 44. checkout compile dataintegration automatic tests metrics collection deploy
- 45.
- 46. data as code the DBA is part of the team
- 47.
- 48. team gets visiblefeedback on builds
- 49. keep the buildgreen
- 50. Wait… I havea big feature!
- 51. branching vs developingon trunk latent code patterns eg. feature flags
- 52. test in acopy of the production env no more “works on my machine”
- 53. today creating vm/containersis cheap! vagrant, docker
- 54.
- 55.
- 56.
- 57. core concept inCD visualize the stages of the process metaphor: assembly line
- 58. Commit Stage Automated Acceptance Stage UAT Manual Testing Capacity Testing Stage Production
- 60. if a stepfails the pipeline stops build binaries once deploy in the same way in every env smoke test deploy deploy in a copy of production
- 61. compile create binaries unit tests code analysis prepares data for subsequent steps Commit Stage
- 62. acceptance tests slowbut high confidence run in parallel whole team responsibility high confidence our software works Automated Acceptance Stage
- 63. what we learnedfrom CI data as code the DBA is part of the team UAT Manual Testing
- 64. infrastructure as code puppet, chef, ansible, docker ops are part of the team UAT Manual Testing
- 65. QA with testers showcase env should resemble production UAT Manual Testing
- 66. You can measure: Scalability Throughput Load Capacity Testing Stage
- 67. deploy is alreadytested on UAT blue/green deploy canary release Production
- 68.
- 69.
- 70. walking skeleton automatebuild automate unit tests automate acceptance test automate release
- 71. So… automation canreduce waste focus on value incremental approach cross functional teams
- 72.
- 73. Questions? https://joind.in/12250 FrancescoTrucchia Vice Presidente @ideato @cphp Michele Orselli CTO @ideato @_orso_