Video: The Future of Software Engineering - Mary Poppendieck
April 7th, 2019
(https://www.youtube.com/watch?v=6K4ljFZWgW8)
[ ] Book: Antifragile - Nicolas Taleb
[ ] Paper: Resilience Engineering: Learning to Embrace Failure - Acm Queue
[ ] Book: The DevOps Handbook - Gene Kim, Jez Humble, Patrick Debois, John Willis
[ ] Book: Continuous Delivery - Jez Humble, David Farley
[ ] Paper: Online Experimentation at Microsoft - Kohavi, Crook, Longbotham
[ ] Paper: Standis Group Study Reported at XP2002 by Jim Johnson, Chairman
[ ] Book: Sprint - Jake Knapp, John Zeratsky, Brad Kowitz
[ ] Video: The Design Sprint - Google I/O 2014
1) scale out
more servers
CAP theorem
less communication
autonomous teams
2) infrastructure as code
conway's law: org structure == architecture
autonomous teams: independent deployment
ops: self-service
there is a cloud in your future
cheaper
more stable
more secure
more expandable
infrastructure as code
containers
serverless = event-driven
software-defined-networks
3) The New Technology Stack
dependency problem
api > db
big data
resilience engineering
Book: Antifragile - Nicolas Taleb
Resilience Engineering: Learning to Embrace Failure - Acm Queue, September 2012
===
Devops
Book: The Devops Handbook
Flow
Feedback
Experimentation & Learning
1) Flow
Book: Continuous Delivery - Jez Humble, David Farley
deploy all the time
release is flipping the switch
2) Feedback
Paper: Online Experimentation at Microsoft - Kohavi, Crook, Longbotham
Paper: Knowledge Discovery & Data Mining - exp-platform.com/expMicrosoft.aspx
Paper: Standis Group Study Reported at XP2002 by Jim Johnson, Chairman
3) Experimentation & Learning
start with signals, not requirements
have a problem statement, not features
plan with hypotheses, not estimates
if you have good problem solving teams, what good are estimates
within constraints
here is my constraints
here is my problem
experiments around a hypothesis, not backlog of stories
analysis and conclusion, not guesses
The hard part is:
What problems are we gonna solve?
Book: Sprint - Jake Knapp
prototype and test any idea in 5 days
before actually coding
Video: Design Sprint Google IO 2014
idea generation (Tuesday)
Conformity Bias
Wednesday: they vote
-> don't like
-> needs majority
-> Attract people to ideas
if you can attract a few ppl to an idea
===
SUMMARY
===
Q&A
Site: www.poppendieck.com
Trend away from monolith and towards distributed systems.
Try and solve problems they don't have.
Over-engineer solutions.
-> Speed of learning
-> Need to migrate
-> Find your own solution
hadoop took 8 years
amazon took 5 years
if you wanna do it in 6 months, you're crazy
What do you think about the role of regulation systems like laws and government in the cloud federated world.
-> there will be regulations
-> cloud ppl better figure it out
-> cloud can be in many different places, and the location where will change rules
This post was referenced in: