Patterns for treating infrastructure as code

London Continuous Delivery Meetup - 9th May 2013

Transcript

1. Patterns for treating infrastructure as code #londoncd
2. @acrmp
3. "Enable the reconstruction of the business from nothing but a source code repository, an application data backup, and bare metal resources." - John Allspaw and Jesse Robbins, Web Operations
4. #1 Include infrastructure code in your pipeline
6. Bring your infrastructure and application pipelines together
7. Application Code Infrastructure Code Base OS
8. Rebuild when other layers change
9. #2 Treat individual modules as software projects
11. Separate code repo
12. Use a tool to manage module dependencies
13. #3 Avoid making manual changes
15. code -> build -> binary
16. code -> converge -> node
17. Reproducible
18. Metrics
19. #4 Aggressively chase fast feedback
21. Parallelise
22. realness -> time ->
23. Well-formedness check
24. Static code analysis
25. Unit Testing
27. Integration Tests
28. Converge and reconverge
29. Lightweight Virtualisation
30. #5 Vendor your dependencies
32. Caching Proxy
33. Build in a off-network environment
34. Schedule Builds
35. Automate population of your artefact repo
36. #6 Define your complete environment in a declarative manifest
38. Use services that expose an API
39. Dumb Orchestration
40. Thanks!