2. AGENDA
I. Organization structures and necessary tensions
II. Common anti-patterns in shifting towards DevOps
III. Tools for a logical rather than structural view
IV. Mapping your organization to a logical view
(the business parts)
V. DevOps Organizational Transition Guidance
VI. DevOps Technical Transition Guidance
3. CONWAY’S LAW
DevOps does not live in a vacuum…
“Organizations which design
systems ... are constrained to produce
designs which are copies of the
communication structures of these
organizations.”
-Melvin Conway (1967)
4.
5. ONE TEAM, ONE PRODUCT
The team discusses and arrives at a decision.
Let’s do
A.
ProductProduct
11
7. TWO TEAMS, ONE PRODUCT
Perhaps some differences of opinion on security or architecture, but
we align on what needs to ship when and get through it.
Let’s do A.
Product 1Product 1 Product 1Product 1
Maybe A’ but
OK
8. THREE TEAMS, TWO PRODUCTS
Let’s do A.
Product 1Product 1
Let’s do B.
Product 2Product 2
Product 1Product 1
Let’s do C.
Product 2Product 2
9. THREE TEAMS, TWO PRODUCTS
Let’s do A.
Product 1Product 1
Let’s do B.
Product 2Product 2
Product 1Product 1
Let’s do C.
Product 2Product 2
14. MATRIXED ORG – DEV VS OPS
$
SLA / Uptime
$
New Feature Delivery
15. DEVOPS ANTI-PATTERNS – THE LONE CONSULTANT
I just met you, yet I fully believe you
have the ability to solve the problems
I've been struggling with for years. My
manager and the product owner have
allocated time for me to both learn new
techniques and contribute as a subject
matter expert to help with the
infrastructure provisioning/deployment
automation. I'm really excited.
I just met you, yet I fully believe you
have the ability to solve the
problems I've been struggling with
for years. I don't have any official
time to work on this, but I'll find
some spare time to help somehow.
I'm really excited.
I’m here
to help!
16. This has been a mess for years. It's
really complicated and you don’t know
what you got into. I’m pretty busy
though and don’t really have time to
waste on this and I’m not paid by the
hour. You’ll be here today and gone
tomorrow.I’m here
to help!
DEVOPS ANTI-PATTERNS – THE LONE CONSULTANT
17. DOAP – THE LONE “DEVOPS” HIRE
I’m here
to help!
This has been a mess for years. It's
really complicated. I’m pretty busy
though and don’t really have time to
waste on this. You’ll see for yourself.
Put in a ticket to request source
control access. Nothing is
documented too well. Maybe you
can make the app compile. Good
luck though.
18. DOAP – THE REBRANDED DEVOPS TEAM
I haven’t done this before
and I’m still doing my day
job as SysAdmin, Release
Engineer, but I’m here to
help!
That’s OK! We’re both behind on
our day jobs, but we’ll find some
spare time and learn and work on
this together! I'm really excited.
19. DOAP – THE PURPOSE BUILT DEVOPS TEAM
We have experience in this
area and time and are here
to help!
That sounds neat. Wait, do I work
with you or IT? IT wanted to do
something else and I don’t have
time to work through this. Why
don’t you figure this out with them?
20. WHY ALL THE ANTI-PATTERNS?
DevOps is not the
language of the rest of
the business.
You can’t just keep
using the word across all
levels of the company
and expect orderly
change.
Need to frame the transformation within
the business context and language
Need a better way to model the
organization
25. COMMANDOS, INFANTRY, AND POLICE
Think of the growth of a company as a military operation, which isn't a stretch, given that both enterprises involve strategy, tactics, supply line,
communication, alliances and manpower.
- Cringely, Robert X. Accidental Empires. Reading, MA: Addison-Wesley, 1996. 235-238. Print.
Commando's parachute behind enemy lines or
quietly crawl ashore at night. Speed is what
commandos live for. They work hard, fast, and
cheap, though often with a low level of
professionalism, which is okay, too, because
professionalism is expensive. Their job is to do
lots of damage with surprise and teamwork,
establishing a beachhead before the enemy is
even aware they exist. They make creativity a
destructive art.
But what they build, while it may look like a
product and work like a product, usually isn't a
product because it still has bugs and major
failings that are beneath the notice of commando
types. Or maybe it works fine but can't be
produced profitably without extensive redesign.
Commandos are useless for this type of work.
They get bored.
Most of business and warfare is conventional. But
without commandos you'd never get on the
beach at all.
Grouping offshore as the commandos do their
work is the second wave of soldiers, the infantry.
These are the people who hit the beach en
masse an slog out the early victory, building the
start given by the commandos. The second wave
troops take the prototype, test it, refine it, make it
manufacturable, write the manuals, market it, and
ideally produce a profit. Because there are so
many more of these soldiers and their duties are
so varied, they require and infrastructure of rules
and procedures for getting things done - all the
stuff that commandos hate. For just this reason,
soldiers of the second wave, while they can work
with the first wave, generally don't trust them,
though the commands don't even notice this fact,
since by this time they are bored and already
looking for the door. While the commandos make
success possible, it's the infantry that makes
success happen.
What happens then is that the commandos and
the infantry advance into new territories,
performing their same jobs again. There is still a
need for a military presence in the territory.
These third wave troops hate change. They aren't
troops at all but police. They want to fuel growth
not by planning more invasions and landing on
more beaches but by adding people and building
economies and empires of scale.
Commandos Infantry Police
Original Publication: 1991
Revised: 1996
26. SPOTIFY GUILD MODEL
Scaling Agile@ Spotify
With Tribes,
Squads,
Chapters
& Guilds
Henrik Kniberg &
Anders Ivarsson
Oct 2012
27. SPOTIFY GUILD MODEL – SELF CONTAINED TEAMS
D – Dev
Q – QA
S – SysAdm / DevOps
p - Pioneer
s - Settler
t - Town Planner
Dp
Ds
Dt
Qp
Qs
Qt
Sp
Ss
St
Dp
Qt
St
Dt
Ds
Ds
Ds
Qp DevOps
Guild
Ds
Ds
Qp
Ds
Ss
SsQs
Dp
SpDevOps
Guild
X,Y
28. SPOTIFY GUILD MODEL – MATRIXED TEAMS
D – Dev
Q – QA
S – SysAdm / DevOps
p - Pioneer
s - Settler
t - Town Planner
Dp
Ds
Dt
Qp
Qs
Qt
Sp
Ss
St
Dp
St
DsDs
Qp
Ds
Qp
Ds Ss
SsQs
Sp
DevOps
Guild
X,Y
IT IT
29. PREPARING THE STRUCTURE PLAN
Overlay virtual teams with
P/S/T exercise to get at an
effective design you need
It contains our desired
mix of product maturity
lifecycle domain
expertise
How do we make it
happen?
We have a desired end
goal virtual team
DevOps Guild layout
30. DEVOPS TEAM TO DEVOPS GUILD
Key permanent core members serve as:
Facilitators
Evangelists
Incubators of environment/deployment/monitoring practices
Curators of established reference models
Recipients of outside training and leaders of internal training
DevOps Team → Tools & Automation Practices Team
Satellite members from dev/QA teams:
SME’s for application internals
Know highest risk areas needing better monitoring/logging
Know definition of success, how to test for it, transition as much as
possible to automatic
31. DEVOPS TEAM IS NOT
Your permanent new release management team
•Manual practice and understanding of deployment WITH Dev and/or IT
is a very short term step to automation
•End Goal: Release responsibilities should be automated and then
owned/initiated by the dev teams
On hand team augmentation for general Dev/IT/SysAdmin
•Temptation to pull people for other tasks/fires can be great
•BUT Some group needs a focus on continuous improvement activities
leading towards faster and safer releases or there won’t ever be less
fires.
Change this maybe not tomorrow, but soon.
32. TRUE CROSS-TEAM EFFORT
Dev needs official time for collaboration on automation and monitoring
QA needs official time for collaboration on automation, data setup, and
automated testing
Long-term program NOT one person’s lone side project
33. FACILITATION IS IMPORTANT!
SIMPLE METAPHOR…
Mgr Email To: Bob, Nancy, Jim, Sally
Subject: Please get X done
Result: Waiting. Some pockets of activity.
Instead…
Mgr Email To: Nancy
CC: Bob, Jim, Sally
Subject: Please get X done with assistance from Bob, Jim, Sally
Result: Everyone is accountable. Nancy will facilitate. Self organize from there.
Given management support, DevOps team can facilitate.
34. FACILITATION IS NOT HEAVY PROJECT MANAGEMENT
1. There should not be comparable time meeting and talking about tasks not
getting done as time spent working on them.
2. Pairing at one computer with someone on another team IS helpful.
3. Start with agreement on shared goals and time commitment. You’ve failed
already without this.
4. Initially work can be entered, estimated, and prioritized in each team’s home
tracking system (JIRA, tickets, etc) rather than adding a side process to each
team. Make any alignment points or deadlines clear to all.
5. Track and celebrate measurable overall DevOps goals progress.
IF (Issues) GOTO 3
6. Revisit cross-team work process as needed.
35. CHALLENGES
Moving from recreational to organized sports often takes
painful unlearning of old techniques and learning new
ones and being worse in between
Setup will get more complex before it gets easier
After better abstractions and automation are in place, it’s
far better than before
- Arthur C. Clarke.Any sufficiently advanced technology
is indistinguishable from magic..
-Arthur C. Clarke
37. WORKING TOGETHER
We need to change the culture, especially in IT operations,
from one of viewing ourselves as ‘order takers’ to one of being
full team members.
-Gene Kim
38. GENE KIM?
We need to change the culture, especially in IT
operations, from one of viewing ourselves as ‘order
takers’ to one of being full team members.
-Gene Kim
Creator and CTO of Tripwire – one of the first file integrity
monitors authorized in PCI DSS
Author of prominent ITIL book
Well respected security community member
DevOps thought leader
Author of prominent THE DevOps book
40. LISTEN AND RE-EVALUATE
3. Can a different
process meet the
need BETTER
than before for all
involved?
2. Which
processes
are
structural
rather than
logical?
1. LISTEN to
goals and
processes of
pre-existing
groups
44. COMMUNICATE, REPEAT, REPEAT, REPEAT, REVIEW
Make it a real
project & hold
people accountable
for doing their part
Review and revise
Communicate plans and expectations
repeatedly all the way down to dev
managers / team leads. Everyone’s
direct boss needs to know the
importance of the work and have time
budgeted. Hard to ask people to spend
time on things without their direct boss
buy-in
45. FINAL TIPS ON ORGANIZATIONAL CHANGE
The structuring and often restructuring of departments
isn’t done by people that are irrational.
Complaining about the structures without making any
attempt to understand the underlying motivation is not
helpful.
The most productive negotiations come from a place of
mutual understanding (people need to feel heard)
Lead not only with what change you need, but also what
aspects the other side will also find agreeable
46. WE TRIED THAT ALREADY…
What was different?
Did any parts get traction at any point?
Did you have budget and time allocated for all required team
members?
Was progress on DevOps initiatives a real project with people
accountable for progress?
Did progress stop before reaching a more user friendly automation
end state?
Did you start with the most complex project before hitting a rhythm
and building experience and reusable work?
50. SECURITY / COMPLIANCE – DEV ACCESS
Dev access to prod servers is a further discussion point.
Ideally it can be arranged, but either way there are techniques to reduce the
need for it for many situations.
51. SECURITY / COMPLIANCE – DEVSECOPS MANIFESTO
Leaning in over Always Saying “No”
Data & Security Science over Fear, Uncertainty and Doubt
Open Contribution & Collaboration over Security-Only Requirements
Consumable Security Services with APIs over Mandated Security Controls & Paperwork
Business Driven Security Scores over Rubber Stamp Security
Red & Blue Team Exploit Testing over Relying on Scans & Theoretical Vulnerabilities
24x7 Proactive Security Monitoring over Reacting after being Informed of an Incident
Shared Threat Intelligence over Keeping Info to Ourselves
Compliance Operations over Clipboards & Checklists
http://www.devsecops.org/
http://rewrite.ca.com/us/articles/devops/integrating-infosec-into-the-daily-work-of-dev-and-ops.html
http://techbeacon.com/detailed-analysis-isacas-10-key-devops-controls
52. EVALUATE, AGREE, & MOVE ON
Both have more in common than
not
Little value starting a new initiative
and teams already diverging on
which they use
55. BOLD END STATE GOAL – DEV LAPTOP
ENVIRONMENT
Aim high!
Set a goal to have an end to end testing
environment on a dev laptop with one
command.
•Any tier >1 server in prod is >1 in laptop
environment
•Local data tier with auto-loaded test data
•Unit and integration tests ready to run
•3rd
party services simplest fake that works
•Not trivial but not as hard as people think
Share this quote BEFORE comments start:
Try replacing "we can't because.."
with "we can't until..." and then
we can when..."
-Kent Beck
57. EVERYTHING IS ABOUT FEEDBACK LOOPS
How fast can you identify an issue if you’re free to tear down, restart, turn on
insane level debug logs ANY time you want?
58. NEXT STEPS
Parallel Streams
1.Set up budget/time for training and post training implementation practice for
DevOps culture soft skills & automation for promising DevOps core team members
2.Start training
A.Socialize ideas here to teams
B.Select a project (ideally with change interested engineers on it)
C.Map it to est. Pioneers, Settlers, Town Planners mix
D.Lay out your DevOps Guild (logical/virtual cross-team group)
E.Identify possible member fits from current DevOps team, dev teams, IT
F.Agree on time, budget for activities
G.Share those with DevOps Guild and collaboratively agree on plan with milestones
fitting in time/budget (highly dependent on existing skillset and ramp up on possibly
new technologies)
H.Start initiative
59. ADDENDUM 1: DEVOPS RESOURCES
DEVOPS WORKS FOR ENTERPRISE
SAP Global IT – Continuous Delivery: From dinosaur to spaceship in 2 years
DevOps Enterprise Summit - http://devopsenterprise.io/
DOES Presentations on YouTube
Puppet Labs State of DevOps 2015
DevOps Days – How not to do DevOps: Confessions of a “thought leader”
60. ADDENDUM 2: DEVOPS RESOURCES
DEVOPS WORKS FOR FINANCIAL COMPANIES
Continuous Delivery - The ING Story: Improving time to market
www.slideshare.net/.../continuous-delivery-the-ing-story-improving-time...
Dec 3, 2014 - Continuous Delivery - The ING Story: Improving time to market with
DevOps and Continuous Delivery. 1. Continuous Delivery - The ING Story ...
From Idea to Implementation: DevOps Enabled Agile ...
devopsenterprise.io/.../from-idea-to-implementation-devops-enabled-agil...
Oct 20, 2015 - Three years into an Agile transformation at Bank of America, an
executive who led the effort will take a look at what his team learned along the ...
Capital One Launches Hygieia Open-Source DevOps ...
www.eweek.com/.../capital-one-launches-hygieia-open-source-devops-d...
Jul 27, 2015 - Capital One's engineering team developed a DevOps dashboard called
Hygieia that the company has open-sourced.
61. ADDENDUM 3: USEFUL TANGENTS TO DEVOPS
John Willis: “Devops and Dr. Deming’s 14 Points”
(Velocity NY 2014)
66. Q&A / THANK YOU!
Ask Questions / Share
Concerns
now and/or later
Email: alecl@alecl.com
Notes de l'éditeur
Collection of experience from
* a 20 person small company in the late 90’s and 2000’s
* a global 30,000 person education technology company leading products with millions of students
* a modern startup
* conversations with many industry leaders
Note: Conway’s Law is over 50 years old
Narrative and context is crucial.
The novels The Phoenix Project or The Goal wouldn’t have been as successful had they been two page blog posts with action items.
Let’s dig in.
In the beginning life was simple…
It’s easy to feel part of one team with one mission if you’re all a few feet apart and on the same product.
Features vs Non-Functional Requirements is a timeless concern.
Tension is not a bad thing. Constructive debate keeps you from going to either extreme. Clear value on both sides.
You have no balance this even in a team of 1. It’s not simply structural to an organization but the core of software delivery itself.
Not just enterprise. This applies to growing small companies or even divisions too…
Entering a new market? Working with a strategic partner?
You can’t have 30+ people in a single standup, fully aligned, working closely and in sync in same area without some further structure.
You can call it teams with team leads, holacracy, whatever. It will be different than one team of 4-7 people.
Word product is overloaded
Replace Product with
API
iOS / Android App
Even Microservice
More products -> potentially divergent business and technology goals and schedules
Now things are complicated…
Things get heated and progress slows. What if you didn’t centralize any infrastructure/operational concerns? The dream of Microservices! Is everything fine? [click]
Maybe… but it has its own risks. [click]
Each team reinvents the wheel. Some probably better than others.
Teams needing to grow or shift have more significant ramp up times
Different services will have different maturity and stability
Less people know the deep details of every backend. Bus factor.
Whether you grew as a matrixed organization or self-sufficient teams, you’re likely considering these tradeoffs [click]
Given prominence of Git, we seem to like decentralized nowadays.
What if we change the words a bit…[click]
Hmm. Well, innovation and freedom sound pretty good still, but standardization is not bad. [click]
Now the top doesn’t sound like a clear winner at all.
We like to not repeat ourselves. We like scale and consistency.[click]
The terms on the top and bottom are all valid expressions of the same forces to balance in an organization.
Depending on which side you’re arguing for you may use one or the other. As situations change you may jump between sides. This is normal.
Difference between Dev and Ops is a four letter word [click]
Traditional operations/security culture wants to minimize risk, but moving too slowly has its own risks of being overtaken by more nimble competitors [click]
Demands increase and systems decay so some non-code change can be inevitable. Nimbleness will help you for many origins of change.
The previous tensions are still applicable and valid regardless of your organizational structure
It just gets more polarized in a matrix.
Grossly simplified view. Certainly sales can suffer due to downtime and overall company revenue will suffer if features aren’t competitive and responsive.
So we’ve set the stage. How does DevOps get unfortunately introduced? [click]
How often do team managers plan for time for assisting with a new outside initiative? ALMOST NEVER!
Employees putting in extra time to help someone new that they don’t know or trust when they’re probably behind? UNLIKELY
This is far more likely. A little demotivation for the new person and not much help.
Oh, and if it’s a single full time hire and not a consultant? Maybe someone with “DevOps” in their title? [click]
Perhaps it’s a full time employee and not a consultant. Your team will likely be slightly nicer.
Maybe the new person will be pointed in a direction or even helped for a couple of hours.
Then it’s back to the grind.
The newest person is the wrong person to re-live all the sins of the past of a project ALONE
The DevOps team is still busy doing manual releases and whatever else they used to do, have received no time or training for their new roles, but they are still expected to save the day.
So this is starting to sound better. There’s at least some area of focus.
But still time and lack of concrete direction issues. [click]
On top of that we’ve made another silo and isolated teams with the same goals leads to confusion, delays, duplication of effort, and even turf wars. [click]
So based on these real examples, I’m clearly leaning towards a direction here now, but before we get there let’s take a step back and look at some tools on how we can be intentional about designing organizational structures and not going totally with gut feel.
Why all the anti-patterns? It’s not from ill intent.
<summarize>
We need a different diagram…
Take a step back…
If you’re planning out an office you don’t start with a physical view first.
Why is a logical view important? It focuses on responsibilities, skills, culture, and goals FIRST
NOT on management reporting structures
You need to agree on a logical view of what people are doing where first.
The same applies to your organization structure and the architect in me loves the ideas of a diagram for intentional organization planning and not just an org chart
We’ll look at some mapping tools from Simon Wardley
Early cloud pioneer and organizational thought leader
Creative Commons Attribution-Share Alike 3.0 License.
Pioneers, Settlers, and Town Planners
It’s a far different context building early products that aren’t monetized where maybe a fail whale is OK compared to having multimillion dollar contracts and SLAs to uphold. You don’t necessarily have a town planner mentality trying out new business models but once significant money is being paid to you for services you better have started thinking that way.
Note how work is moved/stolen forward and comes full circle where the most commoditized items can again be used by pioneers. If you have a great logging infrastructure that’s probably no longer a place of innovation. Pioneers can take that and focus efforts on innovation.
CULTURAL norms in different phases are key on to consider
It’s worth really digging into this slide in more time than we have here so definitely go back to it. If nothing else appreciate the value of each role in a mature and sophisticated organization or product.
Your next step is to align your products with an approximate mix of these 3 roles. These are made up examples and won’t all necessarily be at the same company. [click]
Product 1 – Legacy product bringing in a lot of revenue and still has customer growth. New features are useful and attract customers but not revolutionary changes. High desire for stability and economies of scale. Settlers and Town Planners. [click]
Product 2 – Product is 2-3 years old. Still innovating but bringing in paying customers as well. Pioneers still around. Mostly settler mentality helping move towards town planners. [click]
Product 3 – Released 6 months ago. We have paying customers, but still rapidly evolving. This is our first product so no town planner components to rely on yet internally. Probably some external equivalent SaaS offerings though… [click]
Product 4 – Released 2 months ago MVP spin off from original business idea product. We have some town planner infrastructure to help us out. Still innovating rapidly. Not making money yet. Probablly no settlers needed until we prove it out more. [click]
This classification was enhanced and extended by Wardley, but he recognizes he’s seen it elsewhere first. [click]
Robert X. Cringely in Accidental Empires called the roles Commands, Infrantry, and Policy.
The gist is strikingly similar to Pioneers, Settlers, and Town Planners.
Rather than bastardize this great narrative I left the original. You can read it later.
Spotify has self-contained delivery teams called squads. They are ANTI-MATRIX.
Chapters are cross-team virtual organizations of the SAME role (like QA or Dev for instance) for shared efficiencies in learnings
Guilds are a virtual cross-functional team across roles to increase consistency and context.
Tribes are functionally similar areas where multiple delivery teams work. Generally no more than 100 people in a tribe.
Read their paper and watch their videos. While not always perfect or exactly as planned even at Spotify, it’s worth considering compared to alternatives.
How does this help us? [click]
Yes, there are pioneers in DevOps. It’s not all standardization from the start or forever. You have to try out new things sometime before rolling out all over.
If you have a moderate sized organization you won’t have a single DevOps Guild. [click]
You’ll have a few that serve products with similar maturity needs.
This shows all the sysadmin in IT (plus network engineers/etc., just an example)
You’ll find successful people in your organization are unofficialy doing this through relationships.
Why not do it intentionally rather than have success and prioritization be decided unofficialy by relationships rather than business value?
Here’s where we are so far.[click to end]
First we want to clarify the role of the current DevOps team and perhaps rebrand it.
This may not happen
Side projects are good to explore, but they need to be socialized and operationalized
Management is part of this effort!
They provide time, budget, and cover from distractions.
What happens?[click]
Some activity but not a fast way to solution.
Now in this case who is accountable?[click]
Everyone. They were called out to help, but now someone can facilitate and prompt the group.
Email back and forth to solve a technical challenge get wasteful fast.
Sit together or screenshare.
Founder and creator of one of the early and most successful Unix security audit tools
Author of a prominent ITIL book
What does he think of DevOps?… [click]
He’s one of the people that helped catapult its popularity with the seminal book about it!
It is not incompatible to care about security and embrace DevOps. No one is asking to throw out quality or security. Not everything is a tradeoff. Done right those can improve alongside speed of delivery.
1. LISTEN to goals and processes of pre-existing groups AND business
2. Which existing rules and processes are structural (recall Conway’s Law) rather than logical (no pun intended)?
3. Can a different process meet the need BETTER than before for all involved?
Repeat this cycle periodically
--
For example if you have a sign off gate where a VP needs to approve a release or security review can that be solved better through automated security, performance, and quality checks and delegating to the team itself? What if you’re no longer sharing environments so a bad release won’t affect another team? Why slow down then?
Getting budget isn’t easy [click]
Quick aside on Starbucks
In the mid 2000’s they had a decline in sales.
Brainstormed a way to improve quality and they shut down all stores for 3 hours in 2008 so baristas can be re-trained to make the perfect cup of coffee
It cost them $6 million
Improved experience led to increased sales that made up for the one time loss and kept going [click]
Sometimes you have to stop doing the same thing and take a short term productivity hit to get to the next level and far exceed your ability to deliver future value
Not every initiative or customer feature is critical.
Consider what can wait a bit to ultimately build a better foundation for better productivity in the future?
General security practices like granular rights and audit trails can be MORE effectively implemented with automation
Use your providers monitoring/tagging/etc to its fullest, but for app config use AS LITTLE AS POSSIBLE
Almost all your app config should be driven by a modern configuration platform like Chef or Puppet. Pick ONE. Just ONE.
Adhoc scripts not called through Chef/Puppet should be minimized
What does this get you? [click]
Simplicity and consitency in environments.
Have a Dev environment per team, per version, spun up or down whenever needed.
Have your devs and sysadmins have a local way to troubleshoot and experiment
Plus more [click]
Truly mission critical as in life or death or millions of dollars on the line? DR to another cloud vendor [click]
Hybrid cloud if you have apps in a datacenter (or a company you buy does) [click]
Need to cut costs or be able to negotiate with your cloud vendor? Tell them you literally have apps running in a competitor’s cloud.
I snuck in Dev Laptop before. That’s actually a hugely powerful end goal. [click]
Shared data or services are a pain. Don’t worry about stepping on anyone else. Deal with your own test data.
Imagine the end state and then back track with how to get there.
Don’t use the one most complicated, problematic app as an excuse to not move something forward first
Your personal feedback loop -> up to your team’s feedback loop -> you company’s feedback loop
Imagine how fast a developer can troubleshoot and learn if the whole site is on their laptop
W. Edwards Deming philiosophy of quality AND cost improvement
Goldratt’s theory of constraints
Toyota lean manufacturing
Shu Ha Ri – Japanese martial arts stages of learning to mastery
All brought up for a few years now and well worth your time to become familiar with
You can consider aligning your teams across where in the product evolution they stand.