Scaling Product Hunt from 1 to millions of users

How the Product Hunt team was able to scale our systems from 1 user to millions of users. The Talk is focused on our process and technology choices.

Transcript

1. Radoslav Stankov 04/09/2020
   Scaling Product Hunt
   from 1 to million of
   users
   

   View Slide

2. Radoslav Stankov
   @rstankov
   blog.rstankov.com

   twitter.com/rstankov

   github.com/rstankov

   speakerdeck.com/rstankov
   

   View Slide

3. View Slide

4. View Slide

5. View Slide

6. View Slide

7. How it all started
   

   View Slide

8. An Idea
   t
   !
   

   View Slide

9. Funding
   t
   ⛽
   

   View Slide

10. View Slide

11. Product Hunt + AngelList
    t
    #
    

    View Slide

12. View Slide

13. View Slide

14. View Slide

15. View Slide

16. View Slide

17. View Slide

18. View Slide

19. 2019...
    t
    $
    

    View Slide

20. View Slide

21. View Slide

22. View Slide

23. Max number of engineers we ever had was 8
    

    View Slide

24. image credit: remotehub.io
    

    View Slide

25. 3 phases of a startup
    explore expand extract ...
    

    View Slide

26. 3 phases of a startup
    explore expand extract ...
    I think, we are here %
    

    View Slide

27. View Slide

28. Process
    

    View Slide

29. View Slide

30. View Slide

31. “Process is automatic decisions for trivial
    questions and framework for making decisions
    for all other questions.”
    - Rado
    What is process?
    

    View Slide

32. Our process had changed and continues to
    evolve, and the followings are the constants in
    our process over the years &
    

    View Slide

33. ' single player mode
    ( automation
    ) code review
    * constant delivery
    + maintenance
    

    View Slide

34. Single Player Mode
    

    View Slide

35. View Slide

36. View Slide

37. Project
    ,
    -
    .
    %
    Testing 2 weeks fixed sprints
    t
    /
    0
    1
    Project
    Project
    Project
    Project
    Project
    

    View Slide

38. Automation
    t
    2
    3 linters - Brakeman, Rubocop, ESLint, Prettier
    ) unit & e2e tests - Ruby, JavaScript
    + type safety - TypeScript, GraphQL
    * deployment - CircleCI
    4 infrastructure - Terraform
    5 monitor & alert
    

    View Slide

39. Code review
    t
    4
    

    View Slide

40. Code review
    t
    4
    

    View Slide

41. Code review
    t
    4
    

    View Slide

42. View Slide

43. Feature flags
    t
    6
    

    View Slide

44. OFF
    ON
    User
    Check
    Feature flags
    t
    6
    

    View Slide

45. View Slide

46. 1st day 4th day 1st week ...
    

    View Slide

47. 1st day 4th day 1st week ...
    developer
    

    View Slide

48. 1st day 4th day 1st week ...
    developer admins
    

    View Slide

49. 1st day 4th day 1st week ...
    developer admins beta
    

    View Slide

50. 1st day 4th day 1st week ...
    developer admins beta everybody
    

    View Slide

51. View Slide

52. t
    7
    Happy Friday
    8 fix exceptions
    9 database performance
    : upgrade dependancies
    ; pay technical depth & code migrations
    < bug fixes
    

    View Slide

53. View Slide

54. We are a startup, we experiment, iterate, and
    remove what doesn't work.
    Our code needs to be decoupled, so it is easy to
    add and remove features.
    

    View Slide

55. View Slide

56. = have good defaults
    > have good code organization
    ? make common operations easy
    @ isolate dependencies
    A extensibility and reusability
    

    View Slide

57. 2014 2019
    2015 2016 2017
    Architecture History
    t
    B
    

    View Slide

58. 2014 2019
    2015 2016 2017
    Architecture History
    t
    B
    

    View Slide

59. 2014 2019
    2015 2016 2017
    Architecture History
    t
    B
    

    View Slide

60. 2014 2019
    2015 2016 2017
    Architecture History
    t
    B
    

    View Slide

61. 2014 2019
    2015 2016 2017
    Architecture History
    t
    B
    

    View Slide

62. 2014 2019
    2015 2016 2017
    Architecture History
    t
    B
    

    View Slide

63. 2014 2019
    2015 2016 2017
    Architecture History
    t
    B
    

    View Slide

64. 2014 2019
    2015 2016 2017
    Architecture History
    t
    B
    

    View Slide

65. 2014 2019
    2015 2016 2017
    Architecture History
    t
    B
    

    View Slide

66. 2014 2019
    2015 2016 2017
    Architecture History
    t
    B
    

    View Slide

67. ...we plan 6 months ahead, assuming we will
    double our targets
    

    View Slide

68. View Slide

69. The monolith(s)
    t
    C
    

    View Slide

70. Recap
    

    View Slide

71. View Slide

72. View Slide

73. https://speakerdeck.com/rstankov
    Thanks ,
    

    View Slide

74. View Slide