3 surprises when Continuously Deploying infrastructure

Transcript

1. 3 surprises when Continuously Deploying infrastructure
2. What is Continuously Deployed infrastructure?
3. code change CI running infrastructure
4. Fast feedback is key
5. Get iteration time down
6. < 5 minutes
7. Track cycle time (min, max, median, 95e)
8. Get CI close to the action
9. Eliminate latency
10. Chunk your changes
11. Change one, Test one
12. Make it work Make it right Make it fast
13. Make it work Make it fast Make it right
14. 1. Validate idea 2. Pay off technical debt
15. Consider eliminating gates
16. idea production
17. Staging slows feedback
18. 1. Extra tasks every change 2. Less changes pushed into production 3. Lingering state 4. Higher costs
19. BUT it can provide high quality feedback
20. 1. Your changes must apply faster 2. 80% coverage is good enough
21. Don’t start with a staging environment
22. Find other ways to detect errors
23. Think fast and slow
24. Fast tests
25. Smoke tests
26. < 10 seconds
27. Change one, Test one
28. 1. Change app 2. Change DB 3. Change proxy 4. Test app 5. Test DB 6. Test proxy 4. Test app 5. Test DB 6. Test proxy
29. 1. Change app 2. Test app 3. Change DB 4. Test DB 5. Change proxy 6. Test proxy fail early 1. Change app 2. Test app 3. Change DB 4. Test DB 5. Change proxy 6. Test proxy
30. Make feedback visual
32. Slow tests
33. Full stack Integration tests
34. Things you should test
35. Flow of data
36. Example: - Change DNS record - Check record after TTL
37. Simulated client interactions
38. Example (AP): - Write data to store - Query each node has data within n seconds
39. Hook into monitoring
40. Example: - CI runs slow tests - Monitoring queries CI
41. Ship monitoring with code
42. Expose discoverable endpoints
43. Biggest surprise?
44. Actually possible
45. Push all changes through CI
46. More visibility
47. Faster to change Faster to fix
48. Thank you! Questions?