$30 off During Our Annual Pro Sale. View Details »

Failure Is An Option

Failure Is An Option

Presented at Velocity Santa Clara 2015: http://velocityconf.com/devops-web-performance-2015/public/schedule/detail/41680

Despite our best efforts, we fail. From bugs to delivery failures, product planning problems to complex system failures, even the most experienced people in our industry are accustomed to failure. If we accept that failure is indeed inevitable, and embrace it rather than rail against it, how does that change our approach to work?

I’ll begin by talking about failure and its ensuing and varied costs, and compare our instinctive reactions (slow down) and Etsy’s approach (trust and speed), and what we’ve learned along the way.

I’ll cover the advantages of keeping things fast, and how we use technology to help us do that (and stay that way), and also what we do when things do go wrong (as they inevitably will).

I won’t present a one-size-fits-all Royal Road to Safety, because there isn’t one. Instead, I’ll aim to kickstart conversations you can have with your colleagues about your organisation’s attitude to risk, and what changes you might want to make to cope with your own future failures.

Ian Malpass

May 28, 2015
Tweet

More Decks by Ian Malpass

Other Decks in Technology

Transcript

  1. FAILURE is an option IAN MALPASS · @indec etsy

  2. three TRUTHS

  3. you will create BUGS

  4. you will build the WRONG THING

  5. you will not foresee the UNEXPECTED

  6. failure has a COST

  7. MONEY

  8. TIME

  9. DATA

  10. CUSTOMERS

  11. CREDIBILITY

  12. failure is INEVITABLE

  13. expensive failure is NOT

  14. no BARRIERS

  15. just
 SPEED

  16. just
 TRUST

  17. FLEXIBILITY

  18. DEPLOY DEPLOY DEPLOY DEPLOY DEPLOY

  19. None
  20. small CHUNKS

  21. code
 REVIEWS

  22. None
  23. roll FORWARD

  24. manual TESTING

  25. call SECURITY

  26. automated TESTING

  27. no FLAKING

  28. TRY
 (try again)

  29. None
  30. saved by the
 PRINCESS

  31. watch the GRAPHS

  32. None
  33. watch the LOGS

  34. DARK

  35. feature FLAGS

  36. testing the 1%

  37. testing the 2%

  38. testing the 10%

  39. testing the 50%

  40. testing the 100%

  41. MINIMUM-ISH viable product

  42. STAFF testing

  43. PROTOTYPE

  44. A/B testing

  45. expect the UNEXPECTED

  46. PARTIAL failure

  47. we still FAIL

  48. BLAMELESS

  49. LEARN from failure

  50. FRANK analysis

  51. CELEBRATION

  52. None
  53. ⭐️⭐️⭐️⭐️

  54. push the ENVELOPE

  55. None
  56. over to YOU

  57. QUESTIONS?