Technology that builds Organizations

Transcript

1. Technology that builds
   Organizations
   Or not
   

   View Slide

2. Conway’s Law
   - Company’s software architecture usually reflects its organizational structure
   - Changes in Architecture result in organization changes.
   

   View Slide

3. Knight Capital
   August 2012
   

   View Slide

4. Air Traffic Control @ LA
   September 2004
   

   View Slide

5. Ariane-5
   September 1997
   

   View Slide

6. Disclaimer
   22.8% of stats in this talk will be made on the spot.
   

   View Slide

7. Everytime a city’s population doubles, per capita productivity increases by 15%
   - Geoffrey Watts
   

   View Slide

8. What needs to be Solved?
   - Culture - Structure
   

   View Slide

9. Anti-Fragility
   - BitTorrent
   - Miley Cyrus
   - 2013 VMA
   - Novels
   - 46% of top 100 20th Century novels, had a ban attempt
   - Vaccination
   

   View Slide

10. Strain/Downsides
    

    View Slide

11. You are antifragile to Event(E) intensity E(low) -> E(high), if you are better
    off after the event than before.
    - Nassim Taleb
    

    View Slide

12. Chaos Engineering:
    How much does downtime cost?
    

    View Slide

13. Chaos Engineering
    - Netflix terminates Production Instances to Induce latency
    - HP firmware division breaks code to catch in Buildchain.
    - Facebook has project storm.
    - Dropbox DRT
    - Azure
    

    View Slide

14. Schrodinger’s Backup
    

    View Slide

15. Dependency Inversion
    

    View Slide

16. Dependency Inversion
    - Roles, not titles.
    - Roles are Ephemeral
    - Hire People that perform roles.
    - Role == Responsibility.
    

    View Slide

17. Single Responsibility
    

    View Slide

18. You had one Job!
    

    View Slide

19. Single Responsibility
    - !10x Engineer
    - !Transfer of Fragility
    - Skin in the Game
    

    View Slide

20. Communication
    Don’t Call Me, I’ll call you.
    

    View Slide

21. Synchronous vs Asynchronous
    - Responsibility of Failure.
    - Retry and Backoff.
    - Deadline and Expiry.
    

    View Slide

22. Synchronous vs Asynchronous
    - Problem?
    - Hey I have sent you an Email.
    - Hey I have sent you a Slack message.
    - Waiting on a Bus syndrome.
    

    View Slide

23. Topologies
    

    View Slide

24. Centralized vs Decentralized vs Distributed
    

    View Slide

25. Centralized vs Decentralized
    - Tailor specifics
    - Roles and Redundancy
    - Innovation
    

    View Slide

26. - Lead Time Amount of time it takes for an item.
    - Work In Progress Total number of items in Queue.
    - Throughput Number of items getting done per unit of time.
    Little’s Law
    

    View Slide

27. Holacracy
    

    View Slide

28. CAP Theorem
    

    View Slide

29. CAP Theorem
    Node represents an individual capable of making a choice
    - C Any node who makes a decision, informs other nodes.
    - A Every node can make a decision.
    - P Decision cannot be made without involving every node.
    

    View Slide

30. CAP: A-P Tolerance
    - People make decisions, sometimes different
    - Faster decision making.
    - Cultural Independence.
    - Innovational Independence.
    

    View Slide

31. CAP: C-P Tolerance
    - People need to gain consensus to reach the right decision.
    - Sluggish progress
    - Uniformity
    - Minimal distributed ownership cost
    

    View Slide

32. Principles of Packaging
    

    View Slide

33. Coupling
    - Degree to which a node is directly connected to others.
    - Loose Coupling.
    - Services and Contracts.
    

    View Slide

34. Cohesion
    - Alignment between roles.
    - Implementation and Goal must align.
    - Avoid God-Objects.
    - Do one kind of work.
    

    View Slide

35. Encapsulation
    - Restrict direct access to Components.
    - Implementation details abstraction.
    - SLAs all-the-way.
    - Integrational ease.
    

    View Slide

36. Observability
    

    View Slide

37. I will ________ as measured by ____________
    

    View Slide

38. OKRs
    - Transparency
    - Nested Cadence
    - Decouple Rewards
    - Don’t do it all
    

    View Slide

39. Testing & Refactoring
    

    View Slide

40. Testing & Refactoring
    - http://www.nist.gov/director/prog-ofc/report02-3.pdf (Year 2002)
    - Software bugs are costing the U.S. economy an estimated $59.5 billion each year.
    - Users incurred 64% of the cost and developers 36%.
    - While impossible to Eliminate Bugs it can reduce the costs by a /3rd
    

    View Slide

41. Some bugs are too tough to catch in Production
    

    View Slide

42. Tinkering
    

    View Slide

43. Creativity cannot be Quantified
    - Gmail, Maps, Google Talk, Adsense were built in 20% rule.
    - 80 / 20
    - Story of TrustingSocial SRE
    

    View Slide

44. Linux time-sharing
    - Efficient Managers
    

    View Slide

45. Are we here first?
    - Etsy
    - Netflix
    - Amazon
    - HP
    - Ford
    - Mondragon
    - Buffer
    - Zapier
    - Basecamp
    

    View Slide

46. Zappos
    - Offer a new hire the job and at the same time offer them the opportunity to reject the job offer and
    receive $2,000 for doing so.
    

    View Slide

47. Fordism
    

    View Slide

48. Citations
    Fordism
    Conway’s Laws
    Anti-fragility
    Holacracy
    Little’s Law
    Centralised and decentralised coordination patterns
    

    View Slide

49. Who am I?
    Piyush Verma
    Site Reliability Engineering
    Trusting Social
    Twitter: meson10
    

    View Slide

50. Thank you. Credits
    - Aditya Godbole
    - Nishant Modak
    - Mohan Dutt Parashar
    - Talina Shrotriya
    - Akshat Goyal
    

    View Slide