Upgrade to Pro — share decks privately, control downloads, hide ads and more …

DevOps Traction - Continuous Lifecycle London 2016

DevOps Traction - Continuous Lifecycle London 2016

What if you’re not the CTO and you want to improve your quality, performance and stability?
How do you work with low buy-in, legacy, siloed organizations to break down barriers, blur borders and eliminate constraints?

We cover:
- dealing with the enterprise
- tribes and guilds vs silos and departments
- talking the talk and walking the walk
- trust and building confidence
- easy wins
- low investment tools

4fe11dd39e16a687cc15ea45d2fbb829?s=128

Steve Pereira

May 03, 2016
Tweet

More Decks by Steve Pereira

Other Decks in Technology

Transcript

  1. DevOps Traction @steveElsewhere

  2. Story time: @steveElsewhere

  3. Team Sadness @steveElsewhere • Starts with getting woken up at

    3am • Brent is on vacation • Documentation thats outdated • Try a bunch of diagnostic tools • Try to get a hold of a bunch of teammates or escalate • Fix the problem with a gross monkey patch • Fingerpointing / Lost trust • Project gets delayed another month while you pivot
  4. Team Awesome @steveElsewhere • Entire region of infrastructure failed at

    4am, but nobody got woken up • You spend your entire morning working without interruption • Ship everything after getting a positive code review from a teammate • You attend a blameless post mortem in the afternoon • You celebrate a 20% speed improvement from a recent deploy
  5. What does DevOps have to do with it? @steveElsewhere

  6. Why all the fuss? @steveElsewhere 60x fewer failures 168x faster

    MTTR 30x more frequent deployment 200x shorter lead time
  7. Why all the fuss? @steveElsewhere They don't fear change, they

    don't fear velocity To be successful, you can’t be afraid of what makes you successful
  8. How? @steveElsewhere

  9. @steveElsewhere Continunity How? Holistic continuous improvement

  10. @steveElsewhere How? Empower intrinsic motivation Developers want to produce Ops

    staff want to automate If they don’t, they’re afraid of losing jobs, status, freedom
  11. Ok, so nobody has handed you a DevOps utopia

  12. Change vs. Stability Sharing Measurement Easy Wins @steveElsewhere

  13. Local change + holistic view is powerful @steveElsewhere

  14. Change vs. Stability @steveElsewhere Many companies still follow command and

    control
 difficult to change from the bottom up The enterprise: - Authority - Structure - Incentive
  15. Change vs. Stability Don't get burned. @steveElsewhere That means it’s

    very challenging to implement change, especially when the plan is in motion This is why experimentation, rapid iteration and micro services are important I would recommend not starting in the middle of a project, or doing your own thing as part of a team
  16. The Wall(s) @steveElsewhere OPS PMO QA Change Control Legal Security

  17. Some orgs just don't see themselves as software companies yet

    That's ok. @steveElsewhere
  18. @steveElsewhere How many of us know your company mission statement?

    How many of us know your key performance indicators? How many of us know your bosses key performance indicators? Just as if you were selling from the outside, you need to find out who has decision making authority in your org
  19. Is your company ready? @steveElsewhere

  20. The Golden Rule is a bit shit. @steveElsewhere

  21. Think of The Platinum Rule. @steveElsewhere

  22. Know thy audience @steveElsewhere

  23. @steveElsewhere This is knowledge work Knowledge workers know things their

    boss does not know Knowledge must be communicated tactfully in order to resonate They have different incentives and responsibilities than you They are afraid to admit they don’t know something
  24. @steveElsewhere Ask questions: opinion of lean, agile, past experience Understand

    personality types Understand incentives and goals Understand fear Build a relationship
  25. Know thyself @steveElsewhere

  26. Find your gaps @steveElsewhere devopschecklist.com

  27. Beware DevOps @steveElsewhere

  28. @steveElsewhere Aim for specifics

  29. @steveElsewhere Aim for specifics Target your messaging, address incrementally -

    continuous integration - team member on boarding - high availability - disaster recovery - deployment automation - chatops - centralized logging
  30. Sharing @steveElsewhere

  31. @steveElsewhere Nothing happens without sharing - Share an idea -

    Start a guild in your org - Join a meetup and invite those you want to collaborate with
  32. @steveElsewhere Sharing reduces the unknown Reduces fear Builds trust

  33. Share success to build momentum Share failure to avoid it

    in the future @steveElsewhere
  34. Measurement @steveElsewhere

  35. Measurement @steveElsewhere Deploys per dev per day Deployment lead time

    Mean time to recovery (MTTR)
  36. @steveElsewhere Value stream mapping Change Stabilization Validation Release

  37. @steveElsewhere Code Commit Test Env The pipeline Merge Prod

  38. @steveElsewhere The whole #!

  39. Execution Getting it done @steveElsewhere

  40. Change something observe, hypothesize, implement, measure, repeat @steveElsewhere share share

    share share
  41. Low cost, high ROI The tricks of the trade @steveElsewhere

  42. Cut the crap @steveElsewhere

  43. “There is nothing quite so useless, as doing with great

    efficiency, something that should not be done at all.” Peter Drucker @steveElsewhere
  44. Use low-cost, low/no buy-in tools @steveElsewhere

  45. Whiteboard / Post-it notes Slack / Chat StriderCD / Gitlab

    CI / Circle / Travis Vagrant / Docker Ansible @steveElsewhere
  46. The holy grail:
 Continunity @steveElsewhere j.mp/talkresources

  47. Thanks j.mp/talkresources @steveElsewhere