1. When IT Says No:
How To Create Fast Feature Flow
(Without Breaking Everything)
Gene Kim
IT Revolution Press
March 10, 2012
@RealGeneKim, genek@realgenekim.me
3. About Gene Kim
– Researcher, Author
– Industry:
• Invented and founded Tripwire, CTO (1997-2010)
• Co-author: “Visible Ops Handbook”(2006), “Visible Ops Security” (2008)
• Co-author: “When IT Fails: The Novel,” “The DevOps Cookbook” (Coming
May 2012)
– Things I’ve been researching:
• Benchmarked 1300+ IT organizations to test effectiveness of IT controls vs.
IT performance
• DevOps, Rugged DevOps
• Scoping PCI Cardholder Data Environment
3
@RealGeneKim, genek@realgenekim.me
4. Where Did The High Performers Come
From?
@RealGeneKim, genek@realgenekim.me
5. The Downward Spiral
Operations Sees… Dev Sees…
• Fragile applications are prone to failure • More urgent, date-driven projects
put into the queue
• Long time required to figure out “which
bit got flipped” • Even more fragile code (less secure)
put into production
• Detective control is a salesperson
• More releases have increasingly
• Too much time required to restore service “turbulent installs”
• Too much firefighting and unplanned work • Release cycles lengthen to amortize
• Urgent security rework and remedation “cost of deployments”
• Planned project work cannot complete • Failing bigger deployments more
difficult to diagnose
• Frustrated customers leave
• Most senior and constrained IT ops
• Market share goes down resources have less time to fix
underlying process problems
• Business misses Wall Street commitments
• Ever increasing backlog of work that
• Business makes even larger promises to cold help the business win
Wall Street
• Ever increasing amount of tension
between IT
Ops, Development, Design…
These aren’t IT or Design problems…
These are business problems!
@RealGeneKim, genek@realgenekim.me
6. My Mission: Figure Out How Break The
IT Core Chronic Conflict
• Every IT organization is pressured to
simultaneously:
– Respond more quickly to urgent business needs
– Provide stable, secure and predictable IT service
Words often used to describe process improvement:
“hysterical, irrelevant, bureaucratic, bottleneck, difficult to understand, not aligned
with the business, immature, shrill, perpetually focused on irrelevant technical
minutiae…”
Source: The authors acknowledge Dr. Eliyahu Goldratt, creator of the Theory of Constraints and author of The Goal, has written
extensively on the theory and practice of identifying and resolving core, chronic conflicts.
6
@RealGeneKim, genek@realgenekim.me
7. Good News: It Can Be Done
Bad News: You Can’t Do It Alone
@RealGeneKim, genek@realgenekim.me
28. DevOps: It’s A Real Movement
• I would never do another startup that didn’t
employ DevOps-like principles
• It’s not just startups – it’s happening in the
enterprise and in public sector, too
• I believe working in DevOps environments will
be a necessary skillset 5 years from now
@RealGeneKim, genek@realgenekim.me
29. The Prescriptive DevOps Cookbook
• “DevOps Cookbook” Authors
– Patrick DeBois, Mike Orzen, John
Willis
• Goals
– Codify how to start and finish
DevOps transformations
– How does Development, IT
Operations and Infosec become
dependable partners
– Describe in detail how to replicate
the transformations describe in
“When IT Fails: The Novel”
@RealGeneKim, genek@realgenekim.me
31. The First Way:
Systems Thinking
Dev Ops
@RealGeneKim, genek@realgenekim.me
32. The First Way:
Systems Thinking (Left To Right)
• Never pass defects to downstream work
centers
• Never allow local optimization to create global
degradation
• Increase flow to elevate bottlenecks, reduce
WIP, throttle release of work, reduce batch
sizes
• Eradicate blockages in the flow
@RealGeneKim, genek@realgenekim.me
33. The First Way:
Outcomes
• Determinism in the release process
• Packaging responsibility moves to development
• Consistent Dev, QA, Int, and Staging environments, all
properly built before deployment begins
• Decrease cycle time
– Reduce deployment times from 6 hours to 45 minutes
– Refactor deployment process that had 1300+ steps
spanning 4 weeks
• Faster release cadence
@RealGeneKim, genek@realgenekim.me
35. The Second Way:
Amplify Feedback Loops (Right to Left)
• Expose visual data everyone can see how their
decisions affect the entire system
• Protect the integrity of the entire system of
work, versus completion of tasks
@RealGeneKim, genek@realgenekim.me
36. The Second Way:
Outcomes
• Andon cords that stop the production line
• Eradicating “quick fixes” that circumvent the
process
• Ops user stories are part of the Agile planning
process
• Better build and deployment systems
• More stable environment
• Happier and more productive Ops staff
@RealGeneKim, genek@realgenekim.me
37. The Third Way:
Culture Of Continual Experimentation
And Learning
Dev Ops
@RealGeneKim, genek@realgenekim.me
38. The Third Way:
Culture Of Continual Experimentation
And Learning
• Foster a culture that rewards:
– Experimentation (taking risks) and learning from failure
– Repetition is the prerequisite to mastery
• Why?
– You need a culture that keeps pushing into the danger
zone
– And have the habits that enable you to survive in the
danger zone when you're doing 1000 deploys/hour
@RealGeneKim, genek@realgenekim.me
39. The Third Way:
Outcomes
• 15 minutes/day spent on improving daily work
• Continual reduction of unplanned
• More cycles for planned work
• Projects completed to pay down technical debt and increase
flow
• Elimination of needless complexity
• More resilient code and environments
@RealGeneKim, genek@realgenekim.me
41. Meeting The DevOps Leadership Team
• Typically led by Dev, QA, IT Operations and
Product Management/Design
• Our ultimate goal is to add value at every step
in the flow of work
– See the end-to-end value flow
– Shorten and amplify feedback loops
– Help break silos
(e.g., server, networking, database)
@RealGeneKim, genek@realgenekim.me
42. Definition: Agile Sprints
• The basic unit of development in Agile
Scrums, typically between one week and one
month
• At the end of each sprint, team should have
potentially deliverable product
Aha Moment: shipping product implies not just code –
42
it’s the environment, too!
@RealGeneKim, genek@realgenekim.me
43. Help Dev And Ops Build Code And
Environments
• Dev and Ops work together in Sprint 0 and 1
to create code and environments
– Create environment that Dev deploys into
– Create downstream environments:
QA, Staging, Production
– Create testable migration procedures from Dev all
the way to production
• Integrate Infosec and QA into daily sprint
activities
@RealGeneKim, genek@realgenekim.me
45. Integrate Ops Into Dev
• Embed Ops person into Dev structure
– Describes non-functional requirements, use cases
and stories from Ops
– Responsible for improving “quality at the source”
(e.g., reducing technical debt, fix known problems,
etc.)
– Has special responsibility for pulling the Andon
cord
• No ability to restart service without rebooting
• Configuration settings impossible to find
@RealGeneKim, genek@realgenekim.me
46. Integrate Dev Into Ops
• MobBrowser case study: “Waking up
developers at 3am is a great feedback loop:
defects get fixed very quickly”
• Goal is to get Dev closer to the customer
– Infosec can help determine when it’s too close
(and when SOD is a requirement)
@RealGeneKim, genek@realgenekim.me
47. Keep Shrinking Batch Sizes
• Waterfall projects often have cycle time of one
year
• Sprints have cycle time of 1 or 2 weeks
• When IT Operations work is sufficiently fast
and capable, we may decide to decouple
deployments from sprint boundaries
(e.g., Kanbans)
@RealGeneKim, genek@realgenekim.me
48. Definition: Kanban Board
• Signaling tool to reduce WIP and increase flow
48
@RealGeneKim, genek@realgenekim.me
49. IT Operations Increases Process Rigor
• Standardize deployment
• Standardize how unplanned work is
prosecuted: make it repeatable
• Modify first response: ensure constrained
resources have all data at hand to diagnose
• Elevate preventive activities to reduce
incidents
@RealGeneKim, genek@realgenekim.me
50. Letter to Development
• Seek the downstream effects of your actions
– Unplanned work comes at the expense of planned
work
– Technical debt retards feature throughput
– Environment matters as much as the code
• Allocate time for fault modeling, asking “what
could go wrong?” and implementing
countermeasures
@RealGeneKim, genek@realgenekim.me
51. Letter To QA
• Ensure test plans cover not only code
functionality, but also:
– Suitability of the environment the code runs in
– The end-to-end deployment process
• Help find variance…
– Functionality, performance, configuration
– Duration, wait time and handoff errors, rework, …
@RealGeneKim, genek@realgenekim.me
52. Letter To IT Operations
• “The best way to avoid failure
is to fail constantly”
• Harden the production
environment
• Have scheduled drills to “crash
the data center”
• Create your “chaos monkeys”
to introduce faults into the
system (e.g., randomly kill
processes, take out servers,
etc.)
• Rehearse and improve
responding to unplanned work
– NetFlix: Hardened AWS service
– StackOverflow
– Amazon firedrills (Jesse
Allspaw)
– The Monkey (Mac)
@RealGeneKim, genek@realgenekim.me
53. You Don’t Choose Chaos Monkey…
Chaos Monkey Chooses You
@RealGeneKim, genek@realgenekim.me
54. Letter To Product Management
Lesson: Allocate 20% of Dev cycles to paying down technical
debt
@RealGeneKim, genek@realgenekim.me
55. To Designers
• Help IT Operations codify their work and
requirements into great and ever increasing
library of user stories
• Realize that IT processes are likely the largest
impediment preventing your great ideas from
making it to market
@RealGeneKim, genek@realgenekim.me
56. When IT Fails: The Novel and The
DevOps Cookbook
• Coming in July 2012
• “In the tradition of the best MBA case studies, this
book should be mandatory reading for business and
IT graduates alike.” -Paul Muller, VP Software
Marketing, Hewlett-Packard
• “The greatest IT management book of our
Gene Kim, Tripwire generation.” –Branden Williams, CTO Marketing, RSA
founder, Visible Ops co-author
@RealGeneKim, genek@realgenekim.me
57. When IT Fails: The Novel and The
DevOps Cookbook
• Coming in July 2012
• If you would like the “Top 10 Things You
Need To Know About DevOps,” sample
chapters and updates on the book:
Sign up at http://itrevolution.com
Gene Kim, Tripwire
founder, Visible Ops co-author
Email genek@realgenekim.me
Give me your business card
@RealGeneKim, genek@realgenekim.me
58. To Join The Movement
• Sunday 7pm: The Ginger Man
– http://whenitfails.eventbrite.com
• If you would like the “Top 10 Things You Need To
Know About DevOps,” sample chapters and
updates on the book:
Sign up at http://itrevolution.com
Email genek@realgenekim.me
Give me your business card
@RealGeneKim, genek@realgenekim.me
60. Other Resources
• From the IT Process Institute www.itpi.org
– Both Visible Ops Handbooks
– ITPI IT Controls Performance Study
• Rugged Software by Corman, et al:
http://ruggedsoftware.org
• “Continuous Delivery: Reliable Software
Releases through Build, Test, and
Deployment Automation” by
Humble, Farley
• Follow us…
– @JoshCorman, @RealGeneKim
– mailto:genek@realgenekim.me
– http://realgenekim.me/blog
@RealGeneKim, genek@realgenekim.me
Editor's Notes
How each side Actively impedes the achievement of each other’s goals.
Who are they auditing? IT operations.I love IT operatoins. Why? Because when the developers screw up, the only people who can save the day are the IT operations people. Memory leak? No problem, we’ll do hourly reboots until you figure that out.Who here is from IT operations?Bad day:Not as prepared for the audit as they thoughtSpending 30% of their time scrambling, generating presentation for auditorsOr an outage, and the developer is adamant that they didn’t make the change – they’re saying, “it must be the security guys – they’re always causing outages”Or, there’s 50 systems behind the load balancer, and six systems are acting funny – what different, and who made them differentOr every server is like a snowflake, each having their own personalityWe as Tripwire practitioners can help them make sure changes are made visible, authorized, deployed completely and accurately, find differencesCreate and enforce a culture of change management and causality
Who’s introducing variance? Well, it’s often these guys. Show me a developer who isn’t causing an outage, I’ll show you one who is on vacation.Primary measurement is deploy features quickly – get to market.I’ve worked with two of the five largest Internet companies (Google, Microsoft, Yahoo, AOL, Amazon), and I now believe that the biggest differentiator to great time to market is great operations:Bad day: We do 6 weeks of testing, but deployment still fails. Why? QA environment doesn’t match productionOr there’s a failure in testing, and no one can agree whether it’s a code failure or an environment failureOr changes are made in QA, but no one wrote them down, so they didn’t get replicated downstream in productionBelieve it or not, we as Tripwire practitioners can even help them – make sure environments are available when we need them, that they’re properly configured correctly the first time, document all the changes, replicate them downstream
So who are all these constituencies that we can help, and increase our relevance as Tripwire practitioners and champions?How many people here are in infosec?Goal: protect critical systems and dataSafeguard organizational commitmentsPrevent security breaches, help quickly detect and recover from themBad day: no security standardsNo one is complyingYes, we’re 3 years behind. “Whaddyagonna do about it?”Vs. we (Tripwire owner) can become more relevant and add value by help infosec by leveraging all the configuration guidance out thereMeasure variance between produciton and those known good statesTrust and verify that when management says, we’ve trued up the configurations, they’ve actually done itWhy? Now, more than ever, there are an ever increasing amount of regulatory and contractual requirements to protect systems and data
Tell story of Amazon, Netflix: they care about, availability, securityIt’s not a push, it’s a pull – they’re looking for our help (#1 concern: fear of disintermediation and being marginalized)
[ text ] My personal goal is to prescriptively define 1) what does Dev need to do to become a reliable partner, 2) what does IT Operations need to do to become a realiable partner, and then 3) how do they work together to deliver unbelievable value to the business.Of course, the goal is more than happy coexistence. It’s to replicate the Etsy and LinkedIn stories:Increase the rate of features that we can put into production, while simultaneously maintaining the reliability, stability, security and survivability of the production environment.
