Reducing risk with continuous delivery

Reducing risk with continuous delivery

What is continuous delivery and how does it reduce risk with case studies.

This was a lunch time talk at Pivotal (http://www.meetup.com/Pivotal-Labs-Tech-Talks/)

87cee4ccee0b5f4c442d039a9bd0b432?s=128

Joel Chippindale

November 05, 2014
Tweet

Transcript

  1. Reducing risk with continuous delivery BY JOEL CHIPPINDALE - NOVEMBER

    2014
  2. A story about FutureLearn

  3. How many of you work in continuous delivery environments?

  4. How do you explain the benefits?

  5. What is continuous delivery?

  6. What is continuous deployment?

  7. What is deployment?

  8. All the work required to make a new feature available

    to users
  9. Release based deployment

  10. Process

  11. Benefits • QA have time to check each release before

    it goes live • Marketing can announce releases • Defined process sounds reliable
  12. But what happens when something goes wrong?

  13. • Bugs found in QA can delay ALL features. •

    Features may miss the ‘release window’ • How are bugs fixed that are found AFTER QA? • Scheduled downtime?
  14. Big releases are complicated releases

  15. The more complicated the release the greater the chance that

    something will go wrong
  16. Continuous deployment

  17. Process

  18. • Bugs found in QA can delay ALL features. •

    Features may miss the ‘release window’ • How are bugs fixed that are found AFTER QA? • Scheduled downtime?
  19. Benefits • Features and bug fixes available as soon as

    they are ready • Most deployments can be made with zero downtime
  20. Continuous deployment vs. Continuous delivery

  21. Case study 1: Etsy

  22. None
  23. More than 1 billion dollars worth of sales made on

    Etsy in 2012 Source: Contniuous deployment at Etsy
  24. How often do Etsy deploy code?

  25. 25+ times per day

  26. Who can deploy code at Etsy?

  27. EVERYONE

  28. Automated tests

  29. Fast and easy deployments

  30. Constant monitoring

  31. Continual bug fixing

  32. …and gradual role out of features

  33. Case study 2: gov.uk

  34. None
  35. 50 million visits a month Source: GDS blog

  36. Several deployments a day, thousands since launch

  37. Automated tests

  38. Fast and easy deployments

  39. Constant monitoring

  40. …and smoke tests

  41. Case study 3: FutureLearn

  42. None
  43. 700,000 learners

  44. Average of 3 deployments every working day

  45. Automated tests

  46. Fast and easy deployments

  47. Constant monitoring

  48. Continual QA and bug fixing

  49. …beyond Etsy, gov.uk and FutureLearn

  50. Amazon

  51. Google

  52. Facebook

  53. Continuous delivery works well with

  54. Automated tests

  55. Fast and easy deployments

  56. Constant monitoring

  57. Continual bug fixing

  58. These are good practices for EVERY project

  59. Continuous delivery: Reduce risk by delivering small changes often and

    monitoring feedback
  60. Questions? @joelchippindale joel.chippindale@futurelearn.com about.futurelearn.com/blog/