ASAS 2017 - Infrastructure in a DevOps world

I a time where software is eating the world, we need to optimize for business value. For delivering high quality fast. This change also affects what is needed from infrastructure teams. This talk focuses on what a platform is, what it should deliver, what makes a platform good, and what guiding principles to follow in order to build a good platform that is also very maintainable.

Transcript

1. #ASAS2017 @Avisi_ASAS fb.me/asasconference/
2. Infrastructure in a DevOps world Benny Cornelissen Infrastructure Architect @ Avisi @whieee bennycornelissen
3. I am an Infrastructure Architect at Avisi. Previously, I worked at Xebia, TNT, and some other companies. I have been building infrastructure for over a decade, with focus on automation, resiliency and maintainability. I once built mobile datacenters that were built into shipping containers, and that were located in the desert. I also enjoy Belgian beer, coffee, travelling, and cycling.
4. What is a platform and why do I need one?
5. Software is eating the world
6. Software = Business Value But only if customers can actually use it.
7. Speed matters
8. Optimize for speed and autonomy
9. DevOps
10. DevOps
11. Example: "I need a database"
12. "Can't I just build it myself?"
13. Put responsibilities in the right places
14. Platform as a Service
15. Use what's out there
16. Use what's out there
17. Build your own
18. What makes a platform good?
19. Solve the hard problems
20. The right abstraction
21. It should be easy to operate and maintain
22. A platform that is hard to maintain will become a very expensive piece of tech debt
23. Zero downtime maintenance
24. Building a good platform
26. Run in the cloud?
27. Amazon is better at building datacenters than you are
28. Application Principles
29. Twelve Factor App https://12factor.net/
30. Platform is App-unaware App is Platform-unaware (it just runs containers) (it is just a container)
31. Docker App Isolation, OS Abstraction
32. Infrastructure Principles
33. Everything as Code Immutability Fault Tolerant Auto-scaling & Self-healing Decoupled State Smart Self-Assembly
34. Everything as Code
35. Immutability
37. Decoupled State
38. Don't store anything locally unless you're prepared to lose it
39. Fault Tolerant
40. Losing an instance of something should never result in a failed state
41. Recovering from Failure
42. Autoscaling & Self-Healing
43. Example Recovery of failed node
44. CoreOS CoreOS CoreOS CoreOS Auto Scaling Group App App App Proxy Proxy Proxy Proxy Back end Back end Back end Elastic Load Balancer
45. CoreOS CoreOS CoreOS Auto Scaling Group App App Proxy Proxy Proxy Back end Back end Elastic Load Balancer
46. CoreOS CoreOS CoreOS Auto Scaling Group App App Proxy Proxy Proxy Back end Back end Elastic Load Balancer App Back end
47. CoreOS CoreOS CoreOS Auto Scaling Group App App Proxy Proxy Proxy Back end Back end Elastic Load Balancer App Back end CoreOS Proxy
48. minutes 5 humans needed 0
49. minutes 5 humans needed 0
50. minutes 15 humans needed 1
51. minutes 30 humans needed 1
52. minutes ? humans needed ?
53. Smart Self-Assembly
54. Smart Self-Assembly
55. Everything as Code Immutability Fault Tolerant Auto-scaling & Self-healing Decoupled State Smart Self-Assembly
56. What about maintenance?
57. Immutable container platform?
58. Stateful bits (DBs, storage) Container Cluster A Load balancer (Amazon ELB)
59. Stateful bits (DBs, storage) Container Cluster A Container Cluster B Load balancer (Amazon ELB)
60. Stateful bits (DBs, storage) Container Cluster A Container Cluster B Load balancer (Amazon ELB) Redeploy Apps
61. Stateful bits (DBs, storage) Container Cluster A Container Cluster B Load balancer (Amazon ELB) Redeploy Apps
62. Stateful bits (DBs, storage) Container Cluster A Container Cluster B Load balancer (Amazon ELB) Redeploy Apps
63. Stateful bits (DBs, storage) Container Cluster B Load balancer (Amazon ELB) Redeploy Apps
64. actions for a human 3 30 mins to replace a platform mostly waiting 0 downtime zero. downtime.
65. ### Action 1: Create a second platform instance $ terraform apply ### Action 2: Replicate deployments $ restore-units prd ### Action 3: Redirect traffic to new platform $ terraform apply
66. "If all you have is a hammer, everything looks like a nail..."
67. What am I doing at Avisi?
68. Thank You!