Getting to the post-industrial age of DevOps

Transcript

1. Getting to the post-industrial
   age of DevOps
   

   View Slide

2. Nick Bruun
   github.com/nickbruun
   @nickbruun
   bruun.co
   CTO @ Iconfinder
   Fairly carnivorous
   

   View Slide

3. 1. The good old days
   2. The renaissance man’s DevOps
   3. The industrial revolution of DevOps
   4. Trouble in DevOps paradise
   5. The post-industrial age of DevOps
   

   View Slide

4. The good old days
   

   View Slide

5. Database
   Application
   Web server
   Files
   Ops
   Dev
   

   View Slide

6. … but, then
   something happened
   

   View Slide

7. Scale
   

   View Slide

8. Scale
   

   View Slide

9. “Agile”
   

   View Slide

10. Complex stack
    Web server Database memcached Redis ElasticSearch RabbitMQ
    

    View Slide

11. Leaky abstractions
    Mandatory Spolsky-ism
    

    View Slide

12. The renaissance man’s
    DevOps
    

    View Slide

13. Dev Ops
    

    View Slide

14. DevOps
    

    View Slide

15. “Let’s work together”
    

    View Slide

16. (very often) became
    

    View Slide

17. “We all do everything”
    

    View Slide

18. Architecture
    Development
    QA/testing
    Operations
    System administration
    Database administration
    Full-stack developer h4x0r ninja!11
    

    View Slide

19. Inefficient
    

    View Slide

20. Jack of all trades,
    master of…
    

    View Slide

21. The industrial revolution
    of DevOps
    

    View Slide

22. Automation
    

    View Slide

23. View Slide

24. View Slide

25. Tooling
    

    View Slide

26. View Slide

27. View Slide

28. Trouble in DevOps paradise
    

    View Slide

29. “So… what do you do
    for a living?”
    

    View Slide

30. Servers are still servers
    Databases are still databases
    Infrastructure is still infrastructure
    

    View Slide

31. Servers are still complicated
    Databases are still complicated
    Infrastructure is still complicated
    

    View Slide

32. “I need a PostgreSQL
    database”
    Let’s pretend…
    

    View Slide

33. An Ops person would:
    1. Be grumpy, ask questions
    2. Provision server(s)
    3. $ sudo apt-get install postgresql-server
    4. Set up replication, monitoring, logging and backups
    5. Hand over an IP and a port to the Dev
    The Good Old Days
    

    View Slide

34. Some random person would:
    1. Provision server(s)
    2. Pour over the manuals
    3. $ sudo apt-get install postgresql-server
    4. Set up replication, monitoring, logging and backups
    5. Yay, database!
    The renaissance man’s DevOps
    

    View Slide

35. Some random person will:
    1. Provision server(s)
    2. Find a Chef cookbook or Puppet module for PostgreSQL
    and hopefully monitoring/backup/replication
    3. Run configuration management
    4. Debug why things aren’t working (read the manual)
    5. Yay, database!
    The industrial age of DevOps
    

    View Slide

36. But, what if…
    “The primary went down”
    

    View Slide

37. But, what if…
    “We need a bigger server!”
    

    View Slide

38. Oh dear God…
    

    View Slide

39. The Post-Industrial Age
    of DevOps?
    How do we get to the
    

    View Slide

40. How do we get to do,
    what we do best?
    … or,
    

    View Slide

41. As a developer, I will:
    1. Click a button or run a command to get a
    PostgreSQL server
    2. Get back to work
    The Awesome Way™
    

    View Slide

42. View Slide

43. View Slide

44. Service-ification
    noun the encapsulation of everything as services
    

    View Slide

45. Service-ification
    noun the encapsulation of everything as services
    ≠
    Service oriented Architecture
    

    View Slide

46. Services and
    service providers
    

    View Slide

47. Redis
    as a Service
    PostgreSQL
    as a Service
    ElasticSearch
    as a Service
    Load balancer
    as a service
    Application #1 Application #2 Application #n
    ..
    

    View Slide

48. Application runtime
    as a Service
    Redis
    as a Service
    PostgreSQL
    as a Service
    ElasticSearch
    as a Service
    Load balancer
    as a service
    Application #1 Application #2 Application #n
    ..
    

    View Slide

49. Are we there yet?
    

    View Slide

50. Is running infrastructure
    our core business?
    Let’s ask ourselves…
    

    View Slide

51. View Slide

52. “But…
    X as a service is expensive!”
    

    View Slide

53. “But…
    I don’t trust them!”
    

    View Slide

54. Platforms are hard
    

    View Slide

55. Encapsulation/isolation
    

    View Slide

56. View Slide

57. View Slide

58. Resource management
    

    View Slide

59. View Slide

60. Orchestration
    

    View Slide

61. Kubernetes
    

    View Slide

62. Resource
    management
    Encapsulation/
    isolation
    Orchestration
    

    View Slide

63. View Slide

64. View Slide

65. Flynn
    

    View Slide

66. View Slide

67. Metal
    

    View Slide

68. We have a long way to go
    In other words…
    

    View Slide

69. So, why not go back to the
    good old way?
    

    View Slide

70. Focus
    

    View Slide

71. Security
    

    View Slide

72. Resource utilisation
    $
    

    View Slide

73. Uniformity
    

    View Slide

74. Peace of mind
    

    View Slide

75. Let’s get cracking!
    

    View Slide

76. thx
    

    View Slide