Et si ca tourne mal ? Recettes pour des applications résilientes

Présenté à https://cloudnord.fr/
Quel est le rapport entre la plongée sous-marine et des applications cloud-natives ? A priori, pas grand chose. Pourtant, lors que vous vous trouvez à 80 mètres de profondeur, vous devez mettre en place des outils et des techniques pour assurer une résilience absolue. Ces mêmes techniques peuvent s'appliquer à l'architecture et au développement d'applications résiliences. Durant cette présentation, nous verrons tout d'abord une philosophie à adopter dans votre conception d'applications résilientes. Puis nous verrons un ensemble de techniques à avoir à l'esprit et comment les appliquer concrètement dans vos applications.

Transcript

1. Et si ca tourne mal ?
   Recettes pour des
   applications
   résilientes
   

   View Slide

2. And if it goes wrong?
   Recipes for reliable apps,
   learned while scuba diving
   Christopher MANEU
   Startups Cloud Advocate @ Microsoft
   @cmaneu
   

   View Slide

3. @cmaneu
   

   View Slide

4. @cmaneu
   Technical deep diver (trimix, rebreather, >60m / 200 feet)
   Technical instructor
   Solo diving instructor
   

   View Slide

5. @cmaneu
   Scuba diving
   is a safe sport
   

   View Slide

6. @cmaneu
   What is reliability?
   Recipes for reliable apps
   Reliable application are designed to
   recover gracefully from failures and run in
   a healthy state with no significant
   downtime.
   

   View Slide

7. @cmaneu
   Agenda
   Recipes for reliable apps
   #1
   Before your
   dive
   #2
   Go diving
   #3
   Back to
   the shore
   

   View Slide

8. @cmaneu
   Agenda
   Recipes for reliable apps
   #2
   Go diving
   #3
   Back to
   the shore
   #1
   Before your
   dive
   

   View Slide

9. @cmaneu
   Planning for bad things
   Before your dive | Recipes for reliable apps
   We are even dumber deep
   dive under than at the
   surface.
   “
   “
   

   View Slide

10. @cmaneu
    What-if technique
    Before your dive | Recipes for reliable apps
    Do these steps as a team:
    1. Identify dangers
    2. Evaluate risks
    3. Prepare a contingency plan
    4. Repeat
    What if… Then
    Lost mask Backup mask in pocket
    Leak on air tank during transport Recompute my limits, reduce dive
    Leak on main gaz source underwater Switch to redundant gaz supply, close
    faulty gaz tank.
    

    View Slide

11. From What-If to Failure Mode Analysis
    Before your dive | Recipes for reliable apps
    https://aka.ms/cloudnord-scuba
    

    View Slide

12. @cmaneu
    Check-lists
    Before your dive | Recipes for reliable apps
    

    View Slide

13. Resiliency checklist
    Before your dive | Recipes for reliable apps
    https://aka.ms/cloudnord-scuba
    

    View Slide

14. @cmaneu
    Pinnacle dive
    Before your dive | Recipes for reliable apps
    A dive that’s a peak dive for its participants…
    perhaps the deepest, most complex,
    longest, and toughest in their logbook.
    

    View Slide

15. @cmaneu
    Always diving in your level
    Before your dive | Recipes for reliable apps
    • Experience
    • Depth
    • Duration
    • Decompression limits
    • …
    • Change maximum one parameter / dive
    

    View Slide

16. Before your dive | Recipes for reliable apps
    Usage
    patterns
    Availability
    metrics
    Recovery
    metrics
    Availability
    targets
    MTTR
    MTBF
    RTO
    RPO
    SLA
    SLI
    SLO
    https://aka.ms/cloudnord-scuba
    

    View Slide

17. Before your dive | Recipes for reliable apps
    Cost + complexity
    Availability
    Video delivery, broadcast systems
    ATM transactions, telecommunications systems
    Batch processing, data extraction, transfer, and load jobs
    Internal tools like knowledge management, project tracking
    Online commerce, point of sale
    99%
    99.9%
    99.95%
    99.99%
    99.999%
    

    View Slide

18. @cmaneu
    Agenda
    Recipes for reliable apps
    #1
    Before your
    dive
    #2
    Go diving
    #3
    Back to
    the shore
    

    View Slide

19. @cmaneu
    Streamline your configuration
    Go diving | Recipes for reliable apps
    

    View Slide

20. @cmaneu
    Streamline your configuration
    Go diving | Recipes for reliable apps
    

    View Slide

21. @cmaneu
    Streamline your configuration
    Go diving | Recipes for reliable apps
    

    View Slide

22. Is your architecture streamlined?
    Go diving | Recipes for reliable apps
    I Run Kubernetes,
    I have no issues!
    https://k8s.af/
    

    View Slide

23. View Slide

24. View Slide

25. @cmaneu
    When things go wrong,
    a priority framework
    Go diving | Recipes for reliable apps
    #1
    Breathe
    #2
    Buoyancy
    #3
    Ascent
    

    View Slide

26. A priority framework
    Go diving | Recipes for reliable apps
    T
    D
    O
    D
    A
    R
    ime (it’s also team)
    iagnostics
    ptions
    ecide
    ct / assign
    eview
    

    View Slide

27. Brownout strategies
    Go diving | Recipes for reliable apps
    Which part/features of your
    application you can disable
    during an emergency?
    

    View Slide

28. Brownout strategies: a real-world example
    Go diving | Recipes for reliable apps
    

    View Slide

29. @cmaneu
    Agenda
    Recipes for reliable apps
    #1
    Before your
    dive
    #2
    Go diving
    #3
    Back to
    the shore
    

    View Slide

30. @cmaneu
    We need to talk…
    Back to the shore | Recipes for reliable apps
    You know it was a bad dive
    when no ones talk
    on the boat
    “
    “
    

    View Slide

31. Postmortem
    Back to the shore | Recipes for reliable apps
    Never charge team members
    What needs to be said must be said
    Can use different angles to help postmortem
    - Technical
    - Technological
    - Psychological
    

    View Slide

32. Improve: Fail-over checklist
    Back to the shore | Recipes for reliable apps
    https://aka.ms/cloudnord-scuba
    

    View Slide

33. @cmaneu
    Agenda
    Recipes for reliable apps
    #1
    Prepare your
    dive
    #2
    Go diving
    #3
    Back to
    the shore
    

    View Slide

34. @cmaneu
    Agenda
    Recipes for reliable apps
    #1
    Prepare your
    dive
    #2
    Go diving
    #3
    Back to
    the shore
    What-If
    Checklists
    Limits
    Streamlined
    configuration
    Priority
    framework
    Brownout
    strategy
    Debrief
    Improve
    

    View Slide

35. Architecture Review
    Cost, DevOps, Scalability, Resiliency, and Security.
    https://aka.ms/cloudnord-scuba
    

    View Slide

36. @cmaneu
    4 characteristics of a good tech diver
    Recipes for reliable apps
    Humility
    Doesn’t know everything, know that there is more than one good solution.
    Openness
    Doesn’t fear change, don’t reject new or different ideas on the spot
    Analytical
    Doesn’t accept things because they’re just new. Evaluate with neutrality and realistic POV.
    Competency
    Never stop learning
    

    View Slide

37. @cmaneu
    

    View Slide

38. https://unsplash.com/photos/raalZ
    zjZUXQ
    Photo by Jeremy Bishop on
    Unsplash
    Photo by Denes Kozma on
    Unsplash
    

    View Slide

39. Thanks
    Christopher MANEU
    Startups Cloud Advocate @ Microsoft
    @cmaneu
    Recipes for reliable apps,
    learned while scuba diving
    https://aka.ms/chris/slides
    

    View Slide