Reliability in a Serverless world

Serverless @ Nederlandse Spoorwegen

Transcript

1. ComIT
   Driss Amri
   (AWSome)
   February 2023
   Reliability in a Serverless world
   

   View Slide

2. What is Reliability
   

   View Slide

3. Availability Downtime Per Year Downtime Per Month
   99%
   (“Two Nines”)
   99.9%
   (“Three Nines”)
   99.99%
   (“Four Nines”)
   7,2 hours
   3,65 days
   43,2 minutes
   8,76 hours
   4,32 minutes
   52,6 minutes
   

   View Slide

4. View Slide

5. View Slide

6. is the wrong reliability target for
   pretty much everything
   100%
   

   View Slide

7. Reliability on AWS
   

   View Slide

8. View Slide

9. View Slide

10. View Slide

11. View Slide

12. View Slide

13. View Slide

14. Serverless Reliability
    

    View Slide

15. View Slide

16. Multi-AZ out of the box
    

    View Slide

17. Multi-Region
    

    View Slide

18. Challenges
    ● Protecting downstream services that don’t scale as well
    ● Service Limit Quotas
    ● More granular architectures
    ● Denial Of Wallet
    ● Per Function and Service (mis)configuration
    ● Lots of services to choose from
    

    View Slide

19. Chaos Engineering
    

    View Slide

20. Chaos Engineering is the discipline of
    experimenting on a system
    in order to build confidence in the system’s
    capability to withstand turbulent conditions in
    production.
    

    View Slide

21. Chaos Engineering is the discipline of
    experimenting on a system
    in order to build confidence in the system’s
    capability to withstand turbulent conditions in
    production.
    

    View Slide

22. Chaos Engineering is the discipline of
    experimenting on a system
    in order to build confidence in the system’s
    capability to withstand turbulent conditions in
    production.
    

    View Slide

23. Chaos Engineering is the discipline of
    experimenting on a system
    in order to build confidence in the system’s
    capability to withstand turbulent conditions in
    production.
    

    View Slide

24. View Slide

25. View Slide

26. View Slide

27. View Slide

28. View Slide

29. View Slide

30. Common faults
    ● Network
    ○ Latency
    ○ Bandwidth
    ○ Failure to connect
    ○ 4XX/5XX HTTP
    Response
    ● Resource Exhaustion:
    ○ CPU Stress
    ○ Memory
    ○ Disk Space
    ● Weaknesses
    ○ Error handling
    ○ Timeout values
    ○ Events
    ○ Fallbacks
    ○ Fail overs
    

    View Slide

31. Demo
    

    View Slide

32. Demo
    

    View Slide

33. View Slide

34. View Slide

35. View Slide

36. View Slide

37. View Slide

38. Static stability using Availability Zones
    https://aws.amazon.com/builders-library/static-stability-using-availability-zones/
    Beyond five 9s: Lessons from our highest available data planes
    https://www.youtube.com/watch?v=2L1S0zfnIzo
    Chaos testen voor betrouwbaarheid
    https://nsdigitaal.sharepoint.com/sites/TestenBijNS/SitePages/Chaos-testen-voor-betrouwbaarheid.as
    px?source=https%3A%2F%2Fnsdigitaal.sharepoint.com%2Fsites%2FTestenBijNS
    

    View Slide