The 3 C's of Devops
Nathen Harvey, Chef
Culture >>> Character
Common goal >>> Communication
Continuous delivery >>> Continuous Improvements
Character
Trust
Perseverance
Lift up
Respect
Empathy
Tools - TDD, Automated deployment, git, hack days,
available metrics
Communication
Align objective
Infrastructure as code
Metrics
Sit together! Eat together!
Teach
No Asshole Rule
Tools for convos to include distributed people
"Beware of cruel empathy"
Continuous improvement
Kaizen vs. Kaikau
Con't integration
Short feedback loops
Rotate to other departments
Experience product as a customer
Open retros
Traits of success
personal responsibility/empowerment
failure = learning opp
web-oriented arch
allergy to slowness
addicted to data
TO DO:
same room
whole syst. tech platform
reinforce culture w tech
take responsibility
talk to teammates
Understanding the cost of delay
Key takeaways:
There different types of cost of delay to consider
You CAN calculate cost of delay
Linked in business card example
Get the liars in the same room: establish a consistent understanding of piroritity
Filter for decisions
How to create good filters:
Brainstorm
Ask: Can all of these be met? (Conflict)
Ask: Should all of these be met? (Water down)
If you can't choose, use ritual dissent w/ the goal of communicating
"What problem are we trying to solve? "
The show must go on - agile leadership lessons from a life in theater
by John Krewson
REHEARSAL
Each rehearsal builds on each other
Laughs
Commit
Going half-in makes you look foolish
Commit to values
Have courage
Never give a line reading
Explain context
Ask questions
Provide the environment and support needed to get it done
Yes, tech rehearsal is miserable
"Working software is the primary measure of progress"
Support is a level grind
by Jeff Hackett
Help is an aggressive act!
Crisis - Judgement
Average interruptions
56x/day
IQ falls 10 points
36 email checks/hr
3 minutes focused time before switching tasks
16 minutes to refocus
28% interrupted
Keynote: Best Job Ever
Diana Larson
Do work you love to do
work with purpose
find your tribe
Purpose
5 Whys
Teams need purpose
"Keep the internet free"
Your job in 5 words?
Creating successful MVPs in agile teams
by Melissa Perri
"V"
not about building crappy products
not "Fuck it, ship it"
What are we experimenting around?
What is the smallest viable thing we can do to test it?
Setting up experiments
- What is our goal?
- Define the problem and the customer
- What do we want to learn?
- What's the minimum that we can build to test that?
- How do we measure success?
How can we do this in a week?
- Reflect and define
- Specify
- Build and refine
- Build and refine
- Test
Once you build the MVP and validate your assumption ("It works")...
go back and build the whole thing
Interviews are cheap!
Do customers have this problem?
What do they value in my solution?
Devops
"If your whole world were obvious, waterfall would work great!"
Improving flow (or EQUIVALENTLY lead time) is the primary objective of ops
Find a method that guides "what not to produce"
Local efficiencies must be abolished!
can't change too many things at once
"Become the ono"
Understand current conditions
Understand the goal
Identify assumptions
Experiment
"All about wait time"
Be pragmatic
continuous delivery
by Steve Neeley
Stamp and letter exercise
Change to red
Defect found
Antipatterns
Manually deploying software
Manual testing/configuration
Using different binaries in test and prod
Using long-running (>1 day) branches
Prod and test environments different
Culture not giving a damn
Things to do
"I know I just broke the build, but no tests failed." .... write the tests!
Red tests? Don't push code unless it's fixing the tests.
Fast (10 min to run), comprehensive test suite
Parallelize test suites (multi-thread)
Choose the easiest thing to start automating and get good at it
Put everything in version control
Value stream map deployment: Ask: "What takes the most time?" Automate that.
"Imagine we were to build right now, can we ship?"
Use feature toggles
PO turns on toggle!
Stalking agile success in wilds of enterprise
by George Dinwiddie
"When a measure becomes a target, it ceases to be a good measure."
Vary your measure
Change your perspective
What should be there that you don't see?
ZONE OF AWARENESS
vs.
Zone of disturbance
... and how that relates to metrics
When you ask a question, don't immediately answer. Accept uncertainty.
< 3 interpretations - haven't thought about it enough, jumping to a bad decision
2 is a dilemma, often 1 is a foil to make the other look good
Successful evolutionary change of portfolio management
WYSIATI
Exercise
Focus on A project at a time
Better quality
Faster to production
If we need to cancel a project, let's talk about it.
WIP by conversation
A word on Options and commitment
Options have value
Options expire
Never commit early unless you know why
Never over commit unless you know why
Start conversations at the beginning.
"Portfolio kanban is about choosing the work you don't do"
So what was the ultimate golden nugget?
Don't leave it up to chance.
Create it.