SRE: Self Isolated Edition

Transcript

1. armakuni.com
   Site Reliability Engineering
   Self Isolated Edition
   Billie Thompson
   

   View Slide

2. 2
   How are you keeping
   it together?
   

   View Slide

3. armakuni.com
   Site Reliability Engineering
   Self Isolated Edition
   Billie Thompson
   

   View Slide

4. Introductions
   4
   ● Billie Thompson
   ● Developer
   ● Fav. Animal Capybara
   PurpleBooth
   ArmakuniHQ
   Getting to know you, Getting to know all about you
   

   View Slide

5. View Slide

6. We’ve worked with these people
   6
   

   View Slide

7. 7
   I invented a game
   

   View Slide

8. View Slide

9. 9
   30
   Seconds
   Fast
   Careful
   

   View Slide

10. 10
    I can go faster,
    but I’ll make more mistakes
    

    View Slide

11. 11
    I can go slower,
    but I’ll get fewer in
    

    View Slide

12. 12
    Middle pace is about the best
    

    View Slide

13. 13
    30
    Seconds
    

    View Slide

14. 14
    1 Sprint
    

    View Slide

15. 15
    SRE is an
    implementation
    of DevOps
    

    View Slide

16. 16
    In a few months
    SRE will be nothing
    more than DevOps 2.0
    

    View Slide

17. 17
    I want to talk about
    the bits that will be lost
    

    View Slide

18. 18
    Great Powers
    

    View Slide

19. 19
    The metric you see
    is the metric you move
    

    View Slide

20. 20
    The North Star Metric is the single
    metric that best captures the core
    value that your product delivers to
    customers.
    

    View Slide

21. 21
    Focus on feature delivery
    

    View Slide

22. 22
    Focus on feature delivery
    I delivered 15 story
    points this week!
    My teams just don’t have
    the feature velocity
    

    View Slide

23. 23
    Focus on feature delivery
    Focus on stability
    I delivered 15 story points
    this week!
    My teams just don’t have the
    feature velocity
    

    View Slide

24. armakuni.com
    How do you combat the great
    power of feature delivery
    Self Isolated Edition
    Billie Thompson
    

    View Slide

25. 25
    What attributes does
    a countering force
    need to have
    ● Fast feedback cycle
    ●
    ●
    

    View Slide

26. 26
    What attributes does
    a countering force
    need to have
    ● Fast feedback cycle
    ● Be achievable
    ●
    

    View Slide

27. 27
    What attributes does
    a countering force
    need to have
    ● Fast feedback cycle
    ● Be achievable
    ● Have monetary impact
    

    View Slide

28. 28
    Focus on feature delivery
    Focus on stability
    I delivered 15 story points
    this week!
    My teams just don’t have the
    feature velocity and we don’t
    know
    The Error Budget
    

    View Slide

29. 29
    Errors are Inevitable
    

    View Slide

30. 30
    100% Uptime is Impossible
    

    View Slide

31. 31
    100% Uptime is Impossible
    and undesirable
    

    View Slide

32. 32
    Stability-o-meter
    Never down
    Always down
    

    View Slide

33. Our customers would leave
    us if we had performance in
    this region
    33
    Stability-o-meter
    Never down
    Always down
    

    View Slide

34. If we are here we are taking
    as much risk as our
    customers will tolerate
    34
    Stability-o-meter
    Never down
    Always down
    

    View Slide

35. We need a buffer between
    what we aim for and what
    our customers need
    35
    Stability-o-meter
    Never down
    Always down
    

    View Slide

36. This is our error budget we
    can spend it how we like
    36
    Stability-o-meter
    Never down
    Always down
    

    View Slide

37. Maybe we release a risky
    feature
    37
    Stability-o-meter
    Never down
    Always down
    

    View Slide

38. Maybe we do that DB
    migration
    38
    Stability-o-meter
    Never down
    Always down
    

    View Slide

39. Maybe something goes really
    wrong and we go over
    39
    Stability-o-meter
    Never down
    Always down
    

    View Slide

40. 40
    Stability-o-meter
    Never down
    Always down
    SLO
    Service Level Objective
    An internal team target
    

    View Slide

41. 41
    Stability-o-meter
    Never down
    Always down
    SLA
    Service Level Agreement
    An external promise to
    customers backed up with
    financial penalties
    

    View Slide

42. 42
    Stability-o-meter
    Never down
    Always down
    SLA
    An external promise to
    customers backed up with
    financial penalties
    The units of this chart
    are the SLI
    (Service Level Indicator)
    

    View Slide

43. https://cre.page.link/art-of-slos-slides
    Request / Response Availability
    Latency
    Quality
    Data Processing Coverage
    Correctness
    Freshness
    Throughput
    Storage Throughput
    Latency
    SLI Menu
    

    View Slide

44. 44
    A running organisation
    is a complex system in balance
    

    View Slide

45. 45
    What are the balancing
    forces in your organisation?
    

    View Slide

46. 46
    Are they explicit?
    

    View Slide

47. armakuni.com
    Thank you!
    

    View Slide

48. 48
    Questions?!
    Now or at the end
    More More More!
    I Have Questions?
    

    View Slide