2013/11/01 FlowCon

by Gene Kim on

#flowconsf

Up: Adrian Cockcroft: "Velocity and Volume (Or Speed Wins) @adrianco

  • Surely, this is a new high point in life. @adrianco's 1st slide is Phoenix Project! :)
    https://pbs.twimg.com/media/BX_5gAZCYAAu758.jpg
  • .@adrianco: "There's been a 3 order of magnitude increase in rate of change, transaction rate: cust acq, employee hiring, etc."
  • .@adrianco: "Goal of traditional IT: reliable hw to run stable sw; these assumptions are fading into distance; not valid anymore
  • .@adrianco: "I call the talks I've been doing as 'Baffling-late-adopters-As-A-Service." Haha.
  • .@adrianco: "In short, everyone needs to get on top of Agile, continuous delivery, DevOps in order to compete."
  • .@adrianco: "Scale breaks hardware; speed breaks software; Speed At Scale breaks everything" (haha)
  • .@adrianco: "Incidents: want to avoid 'CNN Moment' (invented by eBay): PR/media impact; High cust svc calls; Affects A/B tests"
  • .@adrianco: "Want single digit PR events per year; & we want no more than 1 order of magnitude more for each successive level
  • .@adrianco: "Game Days are important; Fire drills annoying, but it's nice to see everyone in parking lot (not trapped in bldg)
  • .@adrianco: "YY Incidents mitigated by better tools & architecture
  • .@adrianco: "We can switch all users to different zone; takes 30 minutes
  • .@adrianco: "Web Scale Architecture;
  • .@adrianco: "CIO says 'speed it up!'; why? u'r competing w/other orgs; 'Get inside ur adversaries: OODA loop from Korean War
  • .@adrianco: "OODA loop: Observe (competitive move, customer pain point); Orient (analysis, model hypothesis); Decide; Act
  • .@adrianco: "How fast can u go around feedbck loop? Mainframe era: 1 year cycle;" (OMG, this is awesome. $1M to $100MM cost)
  • .@adrianco: "...duration 1-5 yrs; bet the whole company; cost of failure: bankrupt or bought or outsourced"
  • .@adrianco: "Phase 2: Sun era: 3 months cycle time; not board lvl approval, but CIO level; Cost $10K-$10MM; 3-12 months
  • .@adrianco: "Cost of failure: CIO gets fired; product line gets canceled"
  • .@adrianco: "Phase 3: Commodity Era (think Sun/Dell); 2 week agile train; Cost $10K-$1MM; 2-12 wks; bet a product feature
  • .@adrianco: "Cost of failure: product manager reputation (hahaha); not CIO, CEO or company
  • .@adrianco: "What happened? cheaper, faster
  • .@adrianco: "At Netflix, we built Cloud Native: highly agile & highly available service from ephemeral components"
  • Obs (territory expansion; foreign competition; customer
  • .@adrianco: "Continuous Deployment: no time for handoff to IT; the developers must do it themselves; Freedom & Responsibility
  • .@adrianco: "When developer screws up enough, it's time to find a new developer; our airbags have 6" metal spikes attached
  • .@adrianco: "Dev has root access in Prod; & they have pagerduty -- there's no time to xfer info into someone else's head
  • .@adrianco: "@JDeLuccia: Cars should have a metal spike instead of air bags .. imagine how cautious ppl would drive - @adrianco
  • @thinknow: .@adrianco provids thoughtful historical context on how speed and scale change risks and our responses to them.
  • @moshjeier: Put a 6" spike on steering wheel would cause drivers to be more careful. Need peril in system to maintain it
  • .@adrianco: "At Netflix: our IT is a Cloud API; DEVops automation" (accent on the Dev; we teach Dev to do Ops)
  • .@adrianco: "
  • @apwestgarth: RT @stack72: At netflix, devs get root access to production BUT pagerduty to get woken up when it breaks
  • .@adrianco: "Github all the things; we've improved Cassandra so that we could replace Oracle; external vendors were bottlneck
  • @moshjeier: "We get everything from github, and if it isn't on github we write it and put it on github" @adrianco
  • .@adrianco: "In our world: Decide in OODA loop becomes 'just do it; but share your decision' (no approval needed)" (!!)
  • .@adrianco: "This new type of OODA loop is called Innovation, Big Data (ask good q's), Culture, Cloud" (Wow)
  • .@adrianco: "We do entire OODA loop in hours"
  • .@adrianco: "Now: Cost near zero, variable expense; Duration: hrs to days; Bet: decoupled code push; cost of failure near zero
  • .@adrianco: "Product Mgr owns A/B test setup, self svc; Dev: automated tests, self svc deploy, on call; self svc analytics
  • .@adrianco: "Model: Checkin code/Jenkins; Bake AMI/launch; func/perf test; Prod canary test; Prod red/black push"
  • @JDeLuccia: CFO would love 'continuous innovation' and deployments .. if they knew: Cost near zero; variable expense; rapid ROI; etc..
  • @moshjeier: Continuous Innovation Cycle #flowconsf http://t.co/kABlaXCCVc
  • @moshjeier: Continuous Innovation Cycle #flowconsf http://t.co/kABlaXCCVc
  • @eriksowa: RT @jezhumble: Big data: "I want to quickly answer a question that nobody has asked before" @adrianco
  • .@adrianco: "Future: routinely deploy code during day to Europe site (where it's night), and automatically promote to US
  • .@adrianco: "
  • @JDeLuccia: Continuous Delivery and deployment is not risk excluded events.. rather a/b deploys should be done to allow acceptable failures
  • @moshjeier: Netflix global deployment strategy #flowconsf http://t.co/CFcP7RsHud
  • @JDeLuccia: Canary Signatures - automate testing inflow to allow developers to push to production with conditional testing & live feedback
  • .@adrianco Reading List
    https://pbs.twimg.com/media/BYAAvAKCYAAhROC.jpg
  • .@adrianco: Citing "Release It!, Thinking In Systems, REST API Handbook, Black Swan, Drift Into Failure, Continuous Delivery
  • .@adrianco: "More Books: Everything Is Obvious (Once You Know The Answer), Cloudonomics, Lean Enterprise (coming in Jan)"
  • .@adrianco: "Don't read 'Drift Into Failure' on airplanes; it's all about airplane crashes" (haha)
  • .@adrianco: "Takeaways: 1) speed wins; 2) assume broken; 3) cloud native automation; 4) github is your App Store & resume"
  • @moshjeier: "If anybody in the middle of an outage says 'let's restart everything' just say no" -@adrianco
  • @stack72: "Github is an App Store and your resume - both for devs and companies” @adrianco
  • @SoberBuildEng: Takeaway from @adrianco's talk: Netflix shifted costs to healthier/more productive method, not eliminated them. (And that's OK!)
  • @stack72: The keynote at #flowconsf has helped me understand what new systems I need to build to make things faster
  • @JDeLuccia: If #GitHub is the resume of proof and experience of Dev today; what is #infosec equal? GitHub filters creative resume license
  • @JDeLuccia: "Speed Wins" ... in the market, development, and experiences @adrianco
  • @medellre: RT @stack72: You need to be a different LDAP group to get into the SOX controlled systems at Netflix
  • @adrianco: RT @moshjeier: "We get everything from github, and if it isn't on github we write it and put it on github" @adrianco

Up: Jeff Patton, "I Won't Be Fooled Again: How orgs have evolved to value learning over self-deception" @jeffpatton

  • .@jeffpatton: "Sign of person lying: they say, 'trust me, I know what I'm doing'"
  • .@jeffpatton: "In old days, when Mosaic was a browser; we'd dream up idea, build it, deploy it; and then die"
  • .@jeffpatton: "Citing Derek Sivers: 'if you're not failing, you're not learning'"
  • .@jeffpatton: "I've seen testers become part of teams; instead of just checking stuff, they're making huge contributions
  • @BoobBoo: @RealGeneKim @jeffpatton Trust me I am a consultant is worse & offender should be marched offsite -unless it was meant ironically
  • @cbgrasshopper: "I learned that these tester people DO do things. " - @jeffpatton
  • @moshjeier: "I learned that these tester people do do things" -- @jeffpatton
  • @SoberBuildEng: On fooling ourselves: "[We] don't need testers." -@jeffpatton (See also: ops folks... release engineers... UX peeps...)
  • @cbgrasshopper: Loving the tone of humility and learning at #flowconsf from both @adrianco and @jeffpatton
  • @ProdPad: @thinknow @jeffpatton another similar quote: "If it's uncomfortable, you're growing!" That was @noahkagan, I think... Enjoy #flowconsf!
  • .@jeffpatton: "Bottom up: Utility > Usability > Aesthetics; High utility (craigslist); aestehtics & usability (Segway: looking for use)
  • @fcnewtech: @jeffpatton napkin drawing, user experience has layers, start at the bottom #flowconsf http://t.co/CgO403D61W
  • .@jeffpatton: "High
  • .@jeffpatton: "Bottom up: Utility > Usability > Aesthetics; High utility (craigslist); aestehtics & usability (Segway: looking for use)
  • .@jeffpatton: "
  • @moshjeier: Segway as example of product failure, Police on segway "make me want to shop lift" - @jeffpatton
  • @thinknow: "The reason people say 'don't judge a book by it's cover' is because aesthetics matter." --@jeffpatton @comakewithus
  • .@jeffpatton: "When someone tells Dev, 'they're requirements; no really, they're requirements.', it really means 'Shut up'
  • @tonycollander: @jeffpatton shares entertaining experiences of system development - "Won't be fooled again" at
  • .@jeffpatton: "Changing The World For Dummies" Picture:
    https://pbs.twimg.com/media/BYAGoOCCIAA_Sf5.jpg
  • @dominicad: "Requirements - that's when I learned that word meant shutup." @jeffpatton
  • @moshjeier: "requirements are our bright ideas that solve people's problems" - @jeffpatton
  • .@jeffpatton: "Goal is to Minimize Output, but Maximize Outcomes/Impact"
  • @fcnewtech: Fascinating: 1/4? of audience did not raise hands when asked if they had sat with a customer while the cust used their software
  • @thinknow: "Your job is not to make MORE STUFF, your job is to make LESS STUFF" --@jeffpatton @comakewithus #flowconsf #balancedteam
  • @moshjeier: "No matter what the requirements are, if the outcome is bad, we lose" @jeffpatton
  • .@jeffpatton: "Sharing story on LL Bean & product return sw made lots of angry agents/customers"
  • @thinknow: Something I learned from @jeffpatton: take lots of pictures of your workspace and team. You'll be glad you did.
  • @cbgrasshopper: "You can't get empathy from data" - @jeffpatton
  • .@jeffpatton: "It'd be like telling Jane Goodall: no need to go into the field; bring chimpanzees onsite & interview them
  • .@jeffpatton: "Here's what it looks like when analytics crashes for these traders"
    https://pbs.twimg.com/media/BYAI7psCQAAsfsG.jpg
  • .@jeffpatton: "At eBay, @Cagan resp for 60 prod mgrs at eBay; his guidance: Valuable, Usable, Feasible
    https://pbs.twimg.com/media/BYAJqlJCMAEcxjJ.jpg
  • .@jeffpatton: "Showing awesome @Cagan video: 'most innovation breakthrus come from lead developers"
    https://pbs.twimg.com/media/BYAJ7koCMAAa7EO.jpg
  • @moshjeier: With roles #flowconsf http://t.co/MvTvuVFpXS
  • @gwestr: I feel like engineer founders start with feasible, look for valuable, and hire for usable #flowconsf http://t.co/1WeOxkpQpE
  • @fijiaaron: @jeffpatton "You don't get empathy from data" customer service return line doesn't feel right even if the stats look good
  • @stack72: @grey_area @RealGeneKim @medellre and so effective to not get in the way of productivity
  • .@jeffpatton: "@Cagan: '50-80% sw don't meet objectives; but REAL good prod designers fail more" (subtle, they iterate more)
  • @stack72: “It’s only after delivery that we really understand value” @jeffpatton
  • .@jeffpatton: "Eugene Park: Dir of Prod Mgmt at Edmunds: 'we're right only 2/10 times; need flexible arch so we can test/meas
  • @moshjeier: RT @fijiaaron: most software fails to accomplish it's objective @jeffpatton #flowconsf that's called testing
  • .@jeffpatton: "I'm learning that really great architecture is important to enable scalability/performance AND everything else!
  • .@jeffpatton: "How many ppl here have worked with architecture? How many ppl found it old, horrible?" (Same hands. Haha)
  • .@jeffpatton: "Lesson: Architecture isn't just for scalability/perf, but it is critical to enable speed"
  • .@jeffpatton: "Bill Scott, PayPal:
  • @stack72: To me, a great and scalable architecture is only the WHAT we are building. We need to understand WHY we need to build it
  • @cbgrasshopper: RT @randyshoup: "Airbags are rarely if ever used in cars, so clearly they are useless" ;-) @jeffpatton
  • .@jeffpatton: "Lesson: Engineer first for experimentation; focus performance/scalability later"
  • .@jeffpatton: "Lean Startup Build/Measure/Learn is better than OODA loop, b/c it has 3 steps instead of 4" (haha)@thinknow: @moshjeier It's a variant of the "Keely Triangle" http://t.co/I4aSMKTufi @davcron #flowconsf @MrAlanCooper
  • @cbgrasshopper: Architecture optimized for OODA loop cycle time often has the byproduct of scalability and performance.
  • @trivoca: listening to awesome @jeffpatton talk on Edmunds culture of learning:
    https://pbs.twimg.com/media/BYANN7wCUAAX8aK.jpg
  • .@jeffpatton: depressing quote: "We had misperception that process/practice/methods will produce success; most of the time, it doesn't. It does help us learn"

Up: Tim Sullivan, CEO, Ancestry . com: "Creating Flow: An Executive Perspective" (introduced by John Esser)

  • Sullivan was former CEO of match dot com;
  • Sullivan: "Ancestry: 17 yo; 2.7MM subscribers; 12B records; 55MM family trees; 175MM shared photos/docs; $488MM in 2012
  • Sullivan: "Mission: help everyone discover, preserve, and share their family history"
  • Sullivan: "Constant innovation, speed of innovation is the only thing that matters for driving growth [at Ancestry]"
  • Sullivan: "Even marketing spend [subordinates] to innovation; marketing doesn't grow the market, increase adoption" (neat)
  • Sullivan: "2006: Project Cobalt; this is what we don't do anymore: massive, big risk proj to pivot; long list of ideas/reqs
  • Sullivan: "It was a long project, 6+ mo; we needed big leap fw; lots legacy tech; & that's the way we always did it; impatience
  • Sullivan: "Probs: risk (major revamp of cust experience: 'you just hope it works ok', Match .com rollout at 2am, rollback at 5am
  • Sullivan: "Probs: speed (lack of), little customer feedback (loved 2/3 of it, 1/3 of it they hated), lots of chgs at once
  • Sullivan: "Proj Cobalt was tipping point; 'succesful' but painful; knew we had to do things diff; had to get to market faster
  • Sullivan: "Desired state: faster time to mkt; smaller/faster chgs; better alignment w/customers, parallel dev
  • Sullivan: "Answer was to adopt Agile;
  • Sullivan: "Our 3 year transformation: 9mo (1/2011); 2 years: Agile for the enterprise"
  • Sullivan: "
  • @stack72: I love being able to push code whilst away from the office. Just pushed a change to a puppet module from #flowconsf :)
  • Sullivan: "Architecture, infrastructure,
  • Sullivan: "Big event: US govt release of 1940 census: 1sts time we were afraid that competitors could beat us to mkt w/analysis
  • Sullivan: "Ie, company first to analyze/index 1940 census data would win/lose market share; made us commit to continuous delivery
  • Sullivan: "Lesson: it's all about culture (org commitment to work this way); accelerated chg w/o disruption thru small batches
  • Sullivan: "Commitment to rollout small features, test for cust feedback; no more of chgs taking down all of ancestry .com
  • @adr0sen: Sullivan: Continuous Delivery & DevOps thinking key in @ancestry maintaining market position at release of 1940 Census
  • Sullivan: "More innovation leads to happier company!! In 2006, everyone frustrated at how diff to get anything done" (profound)
  • Sullivan: "Today, there's more energy & enthusiasm for the company; b/c we can do more stuff, and get it done faster"
  • @stack72: Its highly inciteful to hear a CEO talk about a companies movement to continuous delivery
  • Sullivan: "Not there yet: we still haven't paid down all our tech code debt; look at all our properties"
    https://pbs.twimg.com/media/BYAaUmnCIAAQl36.jpg
  • Sullivan: "Q: did going private help/hurt xform?" "A: allowed less focus on near term, but not really; these were right for org
  • Sullivan: "Q: how affected budgeting/planning?" "A: realized couldn't write mktg campaign 6 mo in advance; now part of Agile team
  • Sullivan: "...so in order for Mktg to get heads up, must be embedded in Agile teams; budgeting stayed mostly the same"
  • Sullivan: "

Misc

  • @JDeLuccia: Kick off! #flowcon http://t.co/H4ndRIWkf3
  • RT @JDeLuccia: Cars should have a metal spike instead of air bags .. imagine how cautious ppl would drive - @adrianco
  • RT @moshjeier: Put a 6" spike on steering wheel would cause drivers to be more careful. Need peril in system to maintain it
  • RT @moshjeier: "We get everything from github, and if it isn't on github we write it and put it on github" @adrianco
  • RT @moshjeier: Continuous Innovation Cycle #flowconsf http://t.co/kABlaXCCVc
  • RT @SoberBuildEng: Takeaway from @adrianco's talk: Netflix shifted costs to healthier/more productive method, not eliminated them
  • So says a fellow Unicorn! RT @stack72: The keynote at #flowconsf helped me understand what new systems I need to build to make things faster
  • RT @medellre: RT @stack72: You need to be a different LDAP group to get into the SOX controlled systems at Netflix
  • Haha. RT @BoobBoo: @RealGeneKim @jeffpatton "Trust me. I am a consultant" is worse & offender should be marched offsite
  • RT @fcnewtech: @jeffpatton napkin drawing, user experience has layers, start at the bottom #flowconsf http://t.co/CgO403D61W
  • RT @moshjeier: Segway as example of product failure, Police on segway "make me want to shop lift" - @jeffpatton
  • RT @dominicad: "Requirements - that's when I learned that word meant shutup." @jeffpatton
  • RT @fcnewtech: Fascinating: 1/4? of audience didnt raise hand when asked if they had sat w/customer while using their software
  • RT @thinknow: Something I learned from @jeffpatton: take lots of pictures of your workspace and team. You'll be glad you did.
  • RT @cbgrasshopper: "You can't get empathy from data" - @jeffpatton
  • RT @fijiaaron: @jeffpatton "You don't get empathy from data" cust svc return line doesn't feel right even if the stats look good
  • RT @stack72: “It’s only after delivery that we really understand value” @jeffpatton
  • RT @cbgrasshopper: RT @randyshoup: "Airbags are rarely if ever used in cars, so clearly they are useless" ;-) @jeffpatton
  • @moderntime: #ThoughtWorks' Jez Humble at #FlowconSF http://t.co/DyunwfR2Fh
  • @thinknow: Great to see @jeffpatton and @williampietri at #flowconsf #balancedteam http://t.co/WjhVc5qiiS
  • RT @stack72: I love being able to push code whilst away from the office. Just pushed a change to a puppet module from #flowconsf :)
  • RT @stack72: Its highly inciteful to hear a CEO talk about a companies movement to continuous delivery
  • @thinknow: Six Kanban practices on one page via @dominicad #flowconsf http://t.co/locLO36UG6
  • Sullivan: "We left Proj Cobalt, we knew we had to increase R&D spend to get to where we needed to go"
  • Sullivan: "Q: how would u hv accelerated culture chg?" "A: I understood we were doing Agile, but I never served as visible pusher/sponsor; in hindsight, it was so important/transformative, I would have been more attuned
  • @adr0sen: Sullivan: CD timelines major shift on Marketing strategy - from "6 month launch/campaign" thinking to continuous/agile
  • RT @adr0sen: Sullivan: CD timelines major shift on Mktg strategy - from "6 month launch/campaign" thinking to continuous/agile
  • Sullivan: "Definition of Done: we shoot confetti cannons when it's front of customers and they like it" (Nice)
  • Oh, I think he's answering it, I think. "We're spending XX% on Marketing, and much less than that on R&D; that was wrong; R&D is the Customer Acquisition vehicle for the company. It wasn't social, it wasn't viral; I do recall in a board meeting, here's our 5 year vision, we want Marketing as % of revenue to shrink; they crossed last year, and diverge."
  • Sullivan: "Q: has R&D spend gone up?" "A: Cost of Mktg as function of revenue going down, R&D up; crossed last year" (Wow!)
  • Sullivan: "We realized spending more on Marketing than R&D was wrong; R&D is Customer Acquisition vehicle for the company"
  • It wasn't social, it wasn't viral; I do recall in a board meeting, here's our 5 year vision, we want Marketing as % of revenue to shrink; they crossed last year, and diverge."

Up: Randy Shoup, CTO Kixeye: "Presentation: "The Virtuous Cycle of Velocity: What I Learned About Going Fast at eBay & Google"

Presentation: "The Virtuous Cycle of Velocity: What I Learned About Going Fast at eBay and Google"
Time: Friday 12:00 - 12:30 / Location: Robertson 1
eBay and Google operate some of the largest Internet sites on the planet, and each maintains its leadership through continuous innovation in infrastructure and products. While substantially different in their detailed approaches, both organizations sustain their feature velocity through a combination of technology, people, and process. This session will explore how these large-scale sites do it, and will offer some concrete suggestions on how other organizations -- both large and small -- can do the same.
RANDY SHOUP,
CHIEF TECHNOLOGY OFFICER AT KIXEYE

  • @john_atkinson: Randy Shoup speaking the truth. #flowconsf #hiresmart http://t.co/ZYADsEtWjo
  • @cbgrasshopper: People should KNOW they are valued. @randyshoup
  • .@randyshoup: "
  • @moshjeier: Your environment should treat people as valued.
  • .@randyshoup: "Technolgy: Svcs: small teams, well-defined interface, completely independent, autonomy/responsibility
  • @cbgrasshopper: Teams, like services, should have well-defined interfaces to the rest of the company. @randyshoup
  • .@randyshoup: "Quality Discipline: tests help u go faster; tests have ur back; best investment you can make in your own life
  • .@randyshoup: "Test: confidence to brk things; catch bugs earlier/fail faster; never have time to do it twice (esp when constrained)
  • .@randyshoup: "Automating quality:
  • @stack72: Too True! RT @moshjeier: "We don't have time to do it right" WRONG, You don't have time to do it twice
  • @chemphill: amazon 2 pizza teams: keep teams small enough to be fed by 2 pizzas #flowconsf 1 reason I love working @carbonfive
  • .@randyshoup: "Tooling: make it easy to do the right thing; mocking/testing framewk; monitoring; canarying
  • .@randyshoup: "Quality (reliability/scaling) is Priority 0: Google does this awesome, enabling small teams; Old eBay less so
  • .@randyshoup: "Google's investment in testing pays off; enables tons of small teams to work together, all deploying into prod
  • @stack72: "Virtuous cycle of quality: Testing -> solid foundation -> Confidence -> Faster and better”
  • @cbgrasshopper: Developers: don't report bugs; fix them and submit a pull request / patch WITH TESTS. #flowconsf @randyshoup
  • .@randyshoup: "Culture: Autonomy & Ownership: give people/teams autonomy, hold accountable for their success; keep commitments
  • @eshamow: . @ashedryden at #flowconsf “the financial viability of our country and companies are directly related to how diverse the teams we build."
  • .@randyshoup: "Best teams collab across entire value stream: Engr, Prod, Ops; otherwise, playing strategy games vs solving prob
  • .@randyshoup: "Ex of bad organization games: hiding the ball, CYA, politics
  • .@randyshoup: "Google App Engine team: entire team co-located; everyone on floor was part of 'our team'; warms my heart at every telling; building move to allow teams to centralize/co-locate
  • .@randyshoup: "Quality vs Quantity: hard, esp w/Type A; "no, we need more wood behind 1 arrow." "let's solve 1 grt feature"
  • .@randyshoup: "It's always better to ship 1 great feature than 2 iffy ones;
  • @cbgrasshopper: Do less parallel development; serialize features so you can focus and do them well. @randyshoup
  • .@randyshoup: "Whole user / player experience;
  • .@randyshoup: "Culture: Experimentation: Engineer successes; launch is only 1st step; our games are always being iterated
  • @moshjeier: Think holistically about the full end-to-end experience of the user. UX, functionality, performance, bugs, etc.
  • @schmaus: @randyshoup knows software dev. great listening to his talk at
  • .@randyshoup: "Battlefire was overnight success, but War Commander took 6 mo for us to get gameplay right" (!!)
  • .@randyshoup: "Many small exp sum to big wins: Ebay was great: site speed + machine learned rankings = 4% rev incr (!!)
  • .@randyshoup: "Experimentation win: 4% of $6 billion eBay transactions: $240 million added to bottom line" (!!)
  • .@randyshoup: "Also culture of experimentation: Blame-free post-mortem: here's what happened, what did we learn...
  • .@randyshoup: "Leads to perverse/awesome: SRE teams competing to tell biggest failure stories; made my heart sing"
  • .@randyshoup: (this sounds like awesome monthly Intuit life-preserver ritual. Awesome.)
  • @cbgrasshopper: When something goes wrong, don't ask "What did you do?", ask "What did you learn?" @randyshoup
  • @moshjeier: Virtuous cycle of culture #flowconsf http://t.co/YPb1H2wUnX

Up: Roy Rapoport: "Cloud Operations at Netflix: Optimizing Innovation Speed While Supporting Availability"

Time: Friday 13:30 - 14:00 / Location: Robertson 1
This talk is about the role CORE (Cloud Operations and Reliability Engineering) plays at Netflix and its support of Netflix's core software delivery goals (largely focused on maximizing speed of innovation).
In most organizations, groups like CORE end up playing either a gatekeeping role to changes in production (slowing down innovation and being in conflict with continuous delivery approaches) or a manual, runbook-oriented approach to problem solving (which externalizes the costs of, for example, not having an automated deployment pipeline, not to mention the risk of rolling out changes of insufficient quality). At Netflix, CORE's role is as a high-value enabling group focusing on making developers faster and both minimizing and resolving production outages.

ROY RAPOPORT,
MANAGER, MONITORING ENGINEERING AT NETFLIX

  • Rapoport: "Netflix most important: 1) speed of innovation; 2) availability" "we'll get beaton personalization, not uptime"
  • Rapoport: "Freedom/responsibility; Hire smart experienced people (there's not a lot of supervision); Let them loose; Watch magic
  • Rapoport: "What Operations Shouldn't Do:
  • @filler: “Promote freedom and responsibility. Hire smart experienced people. Set them loose. Watch magic happen.” @royrapoport
  • .@royrapoport: "There's really no such thing as 'separation of duty'!" (ie, it's a control, not a control objective)
  • .@royrapoport: "Sep of duty externalizes cost (a terrible thing), removes feedback loops; put smart ppl in dumb systems = dumb ppl
  • .@royrapoport: "Manage by runbook externalizes costs, makes you stupid; stopping people making mistakes is terrible, too
  • .@royrapoport: "Same controls that prevent mistakes often also prevents people from making the right call, too
  • .@royrapoport: "@filler: “Ops shouldn’t deploy engineer’s code. Seperation of duties is not a thing. It’s not in SOX, ISO.” @royrapoport
  • .@royrapoport: "Instead, as Ops, we should: educate Dev (evangelize), help recover faster, break more often, TTD/TTR
  • .@royrapoport: "Ops should publish best practices & availability patterns; build allerting;
  • @moshjeier: truth RT @davemangot: You can't put smart people in a dumb organization to make the organization smarter @royrapoport
  • @SoberBuildEng: "Vendor said: we are going to create our tools so that operators can't violate our policies." -@royrapoport How very Airbus…
  • .@royrapoport: "Our cloud ops is probably in top 5% of people at Netflix about getting value from our alerting platform
  • .@royrapoport: "First question in prod outage: ask 'what changed', but chg ctrl is so hard: Chronos creates single timeline
  • .@royrapoport: "Chronos has massively reduced our time to react and time to recover"
  • .@royrapoport: "Central Alert Gateway: wrote it to augment PagerDuty
  • .@royrapoport: "Chaos Gorilla: simulates loss of Availability Zone; succ to chaos monkey; Chaos Kong simulates loss of Region
  • .@royrapoport: "Ops coordinates firefighting: low-level coordination (not watch dashboards);
  • .@royrapoport: "Ops: Engr Crisis Mgr: formerly would rotate among Dev managers for 3mos, but skills degraded; now dedicated
  • .@royrapoport: "Ops runs Incident Reviews: 'let's talk thru timeline; self-reflection'; but, human error is not a root cause
  • .@royrapoport: "Incident reviews: human error is not a root cause; instead, focus on tools to help prevent
  • .@royrapoport: "What does it take to be in Ops? Grace under pressure, grt tech skills, passion to mk things better, persuasion
  • .@royrapoport: "
  • @SoberBuildEng: .@royrapoport defines firefighting as low-level coordination, crisis management, and incident review. Good, comprehensive def'n.
  • @filler: “If your ops peoples job is to watch dashboards, you’re doing it wrong. Given enough hours, their brains will rot.” @royrapoport
  • .@royrapoport: "Persuasion is important, b/c we can't actually tell people what to do (b/c Freedom & Responsibility values)
  • .@royrapoport: "1) Stay out of the way; 2) help others stay out of the way; 3)
  • @ryantomlinson: RT @stack72: Netflix have a no jerk policy when hiring

Up: John Simone, Heroku: "Designing Continuous Delivery Into Your Platform"

  • @moshjeier: The easy part: version your code, control your deployments, deploy behind feature flags, favor slow rollouts
  • @dylanized: RT @thinknow: "Feedback enables steering" @testobsessed @pivotallabs
  • @gwestr: @j_simone holding court at #flowconsf @heroku @SalesforceEng http://t.co/B6yK12VyFt
  • @moshjeier: The easy part: version your code, control your deployments, deploy behind feature flags, favor slow rollouts
  • @moshjeier: Tricky parts: Version your config, planning db migration (schema, etc), have ability to rollback deployments, canary deployments
  • .@j_simone: "Hard parts: version your environments; stateless & disposable app containers; no snowflake servers

  • @eshamow: .@testobsessed quoting Fifth Discipline - long feedback cycles are like trying to adjust a slow hotel shower temperature dial.

  • .@j_simone: "What id a dev platform? How code gets integrated/deployed; how envs are managed; containers ur code runs in

  • .@j_simone: "..dev platform: ..and the operational data pipeline

  • @moshjeier: Development platform? how code is integrated/deplyd, how environments managed, container code runs in, operational data pipeline

  • .@j_simone: "App versioning: ensure app deployments ONLY happen during CI; 100% repeatable deploys; method of rollback builtin

  • .@j_simone: "Environment: (KEY) disposable containers for code exec (AWS: AMI+setup, Heroku: dyno); recreate on deploy/restart

  • .@j_simone: "recreate upon deploy/restart ensures that it always start from known state; prerequisite for 'disposable'

  • .@j_simone: "More on 'disposable': if containers are long-lived and never die, it's not disposable;

  • .@j_simone: "Origin state only: everything else is 'ephemeral'

  • .@j_simone: "Configs: versioning; built into container & immutable (Heroku does via ENV vars); changing it means deploy

  • .@j_simone: "Configs: feature flags are done at application layer

  • .@j_simone: "Putting it together: visibility of chgs; visibility of curr state; deal w/apps not infrastructure

  • @moshjeier: Feature flagging has to live in the db/api/service at application layer, can't be done via config

Some other awesome talk: Lane's track

  • @thinknow: Why are feedback cycles important? You want a to quickly see if intentions match implementation @testobsessed @pivotallabs
  • @thinknow: "A piece of information today is worth more than that same information tomorrow" @testobsessed @pivotallabs
  • @SoberBuildEng: .@testobsessed: If your agile methodology has a "stabilization sprint," you're effectively fudging the meaning of done.
  • @thinknow: We need FULL CYCLE feedback during #agile cycles. SW quality all the way through customer opinion. @testobsessed @pivotallabs
  • @medellre: RT @fijiaaron: @testobsessed rapid iteration doesn't help without feedback.

@lisacrispin: RT @dominicad: #flowconsf has the distinction of 1st tech conference where I could run into ppl I needed to talk to in the ladies room.

Up: James Deluccia, E&Y: "Audits Of High Deployment Environments", @jdeluccia, an auditor

  • .@jdeluccia: "Role of assurance: think "OAuth" for business; (haha); enable businesses to trust each other; 3rd party is audit
  • .@jdeluccia: "Auditors (those issuing the reports) stake their names & hv liability w/each issuance; think SOX & financial stmt
  • .@jdeluccia: "Ex: u're working w/tech partner to provide JIT equipment: must know about supply chain risks; is ops consistent?
  • .@jdeluccia: "..are they secure? why? prev supplier had competitor hack their shipping manifests; ergo, can't rely on partner
  • .@jdeluccia: "this led to a much more sophisticated supply chain management process, with higher assurance"
  • .@jdeluccia: "In most orgs, u audit yourself (eg, continuous improvement); in regulated industries, reports go to externally
  • .@jdeluccia: "survey findings: more vendors & clients mandating reports; 45% of cases, audit reports don't meet objectives
  • .@jdeluccia: "For every audit objective, it takes on average 40 hours to respond" (!!)
  • .@jdeluccia: "Regulations & other 3rd parties xfer risk to your business & ask for proof; thus comes 'control objectives'
  • .@jdeluccia: "Ex: control objectives: 'dev, testing & ops environment should be separated to reduce risk unauthorized access'
  • .@jdeluccia: "When auditor asks, show me all code pushes are authorized; ur goal is to show evid that the ur process is working
  • .@jdeluccia: "A bad day for auditor is when someone shows you 200 printed pages of logs"
  • .@jdeluccia: "What auditors need: demonstrative evidence that 'control objective' is satisfied to a statistically relevant level
  • .@jdeluccia: "Q: out of 100Ks of deploys, here are logs; procedures were followed: here are reds & greens; we're in compliance
  • .@jdeluccia: "A: It depends; often, it's that easy; caution: is population really representative of the risk
  • .@jdeluccia: "Reach out to auditors 1 yr in advance; Audit programs are built 1y in advance; built upon last year's risk
  • .@jdeluccia: "As a team, ensure you know what regulatory & common control obj that business is managing against (PCI, SOX
  • .@jdeluccia: "Ask auditors: what is consolidated control objectives that I need to achieve, then get signoff (rt abstraction level)
  • .@jdeluccia: "Once you understand, say, the 17 control objectives; now you can work w/teams to describe form of evidence of effectiveness
    • Examples of 17 control objs: access controls, management signed off, changes appropriately managed, plan
  • .@jdeluccia: "Require auditors to come back with feedback; shows that you have continual improvement in place"
  • .@jdeluccia: "Change management: tier out different types of chg risks; responsive control structure; have magical testing middle management; you want credit for it.
  • .@jdeluccia: "On chg mgmt: You need to get credit for all the automated testing; you're NOT just pushing into PRODUCTION!"
  • .@jdeluccia: "On automated testing; all the tests serves as this invisible, magical box in the middle between Dev and Prod"
  • .@jdeluccia: "PCAOB
  • .@jdeluccia: "How to help audtiors? Clarity, 100% awareness of population (auditors need to know u'r doing exactly what u need to do)
  • .@jdeluccia: "How to help? Multiple compentiencies involved, reproducible, comparable, integrity/validity
  • @JDeLuccia: 6 servers deployed; 4 w/ wrong version & 900 "issues" in one category of systems at http://t.co/3Ux9dabr4H http://t.co/gE3dvdoGiS
  • @JDeLuccia: 6 servers deployed; 4 w/ wrong version & 900 "issues" in one category of systems at http://t.co/3Ux9dabr4H http://t.co/gE3dvdoGiS

Sascha

  • @SoberBuildEng: "I have a real passion for package management... BECAUSE I HATE IT!"-@sascha_d
  • @filler: “People want prod outside of prod, but I always end up finding that dev kickstart server, untouched, getting no love.” @sascha_d
  • @filler: “It’s much more fun to write pipeline code than it is to run pipeline scripts.” @sascha_d #flowconsf #realopstalk
  • @filler: “Eliminate mistakes. Get people out of the pipeline.” @sascha_d
  • @stack72: The more manual steps you have in your environment, the more mistakes we can introduce
  • @filler: Ops should most definitely own ‘environments’. Not the particular env per se but provide a trustable, flexible, rebuildable env.
  • @randyshoup: "Dev [environment] is production for developers"; they spend their days and nights there @sascha_d
  • @liamjbennett: “@stack72 you can’t keep making the release engineer figure things out - they will leave #flowconsf” - :-)

Mainframe

  • @eshamow: RT @moshjeier: @eshamow Mainframe developers are in huge demand due to the large # of retiring employees@cbgrasshopper: RT @eshamow: Incredibly impressed by the talks at #flowconsf thus far. More than anything else it's about systems thinking in the trenches.
  • @BMC_DevOps: RT @moshjeier: How to make a large multi-million $ project successful. #flowconsf http://t.co/sACkTzvQQY
  • @moshjeier: How to make a large multi-million $ project successful. #flowconsf http://t.co/sACkTzvQQY
  • @BMC_DevOps: "business cases that read like a Hercule Puerot novel and I don't know who did it"

@j
* @adr0sen: It's estimated ~200 billions line of COBOL code runs in production systems around the world today. Wow
*

Up: Catherine Courage, SVP Customer Experience, Citrix: "Inspiring Design Thinking In Corporate Culture"

  • Courage: "Citrix mission: enable mobile workforce"
  • Courage: "10 yrs ago, everyone wanted more features, faster; that doesn't work anymore"
    https://pbs.twimg.com/media/BYBUtjTCMAAVJtL.jpg
  • Courage: "Exemplars of user experience: Apple, Target, Virgin Air, Disney, Tesla, Ritz-Carlton, IKEA
  • @stack72: Session for all parts of the spectrum at #flowconsf - should have been called systems thinking conf :)
  • @moshjeier: RT @pkdotnet: "The Dev environment is the production of developers." Give it the effort it deserves. @sasha_d
  • Courage: "Design drives entire experience, every customer touch point"
  • Courage: "If you're in small company, I urge you to try to make Design Culture; In large company, still try. :)"
  • Courage: "UX not icons; it's relatable, credible, visible (@cagain)
  • Courage: "Everyone has customer, even things like HR: HR customer? Employees"
  • @thinknow: Make Design part of the company DNA. Think "what is the value to the customer?" (it's not "reduce call volume!") @ccourage
  • Courage: "
  • @moshjeier: Design isn't just for UX and designers, everyone has a role in making design matter. It's not just buttons, colors and layouts
  • @stack72: Design is not just buttons, colours and layout
  • @filler: "Reduced call volume is a bad metric. Can achieve that by burying the phone #. Its about added measured customer value" @ccourage
  • @moshjeier: Framing problems is really important: make the customer the focal point
  • @moshjeier: Talking about @VirginAmerica being a great customer experience. They are also @salesforce customers :)
  • @moshjeier: I'd love to meet during drinks session. Your tweets rock. :)
  • @stack72: One of the design principles at Citrix is to inspire delight :)
  • @cbgrasshopper: RT @stack72: Companies that seem to get the WHY being discussed at #flowconsf http://t.co/D85HdTiwR6
  • @filler: "Our 5 goals:

Make it simple.
Exhibit craftsmanship.
Inspire delight.
Deliver unique value.
Focus on human goals."

@ccourage
* .@ccourage: "
* @moshjeier: This is definitely a trend during #flowconsf "Walk a mile in their shoes." Customers are why we get paid, let's not forget them.
* .@ccourage: "Since 2011: 47 classes/workshops taught; many locations
* .@ccourage: "Credible: how do we show design impact? here's what consoles looked like: 20y old"

https://pbs.twimg.com/media/BYBXndMCQAAdeat.jpg
* .@ccourage: "
* @filler: Once Citrix employees buy into design principles, value design, they have internal courses to keep up the inertia. @ccourage
* @moshjeier: RT @stack72: You need to be able to demonstrate what your company talks about. Otherwise you loose credibility
* @davemangot: Simple doesn't mean simplistic @ccourage
* .@ccourage: "Experimenting; now we don't start w/solution in mind; instead, rapid prototype, storytelling, experiments
* @stack72: This talk really demonstrates that Citrix are trying to find their WHY. They are making sure their employees are buying into it
* .@ccourage: Awesome. "Iterate iterate iterate iterate... UNTIL AWESOME. It'll be ok. :)"
https://pbs.twimg.com/media/BYBYaO9CEAE3691.jpg
* .@ccourage: "Michael Jordan: I've missed 9000 shots; I've lost 300 games; 26 times, I was trusted to take winning shot & missed
* .@ccourage: "[For creative work], I hate cubicles & boardroom tbls; I never want to see one again; I want to see collaboration"
* @moshjeier: Call out the people that go above and beyond for their customers. "Design Heroes" @ccourage
* @stack72: RT @moshjeier: Get people out of cubicles and board rooms. Get people out into the open, make the collaboration visible.
* .@ccourage: "I need to show execs, ppl at top the value of design culture, using their language & culture (eg: ROI, visuals)
* .@ccourage: "Never wait to be told to something; if it can impact customers, just dig in & do it; ask for forgiveness later
* @chemphill: citrix circa 2010: we have a design team? great, I need an icon @ccourage
* @mblongii: The most common form of entropy in an organization is apathy about the mission.
* @filler: Regarding detractors and pessimists, "You will only be as good as the team you have around you." @ccourage
* Love it; execs think customer case studies are the success stories you post on website; instead, I'm bringing stories about how Sue hates our product

  • TweetScriber; we wrote it; it's free; >>> it's all about having awesome friends. :)

Up: Linda Rising, "Organizational Change Myths and Patterns for Evangelists"

Time: Friday 16:40 - 17:25 / Location: Fisher Room
You have great ideas. You're smart. The people in your organization are smart. Post Hoc, Ergo Propter Hoc -- the rest is easy -- simply make a transition plan and transparently explain the benefits. These are other organizational change myths will be tackled by Linda's talk about patterns for introducing new ideas. She will provide some useful tips for helping you start on Monday morning to grow, step by step, any innovation.

LINDA RISING,
QUEEN OF PATTERNS. AUTHOR OF NUMEROUS BOOKS

  • @BMC_DevOps: #flowconsf @realgenekim trying @adrianco google glass http://t.co/Fs765MhLyc

  • @filler: "Now you have some new ideas, and you are wondering 'What can I do on Monday morning?' Perfect. Thats why I'm here." @RisingLinda

  • .@RisingLinda: "

  • @moshjeier: Giving a name to a recurring problem with a known solution means the names of rltd ptrns can be used to talk the probs and sltns

  • .@RisingLinda wrote book "Fearless Change"; Ph.D. in computer sci; patterns based on social psychology, influence strategies

  • .@RisingLinda: "I went back to school many times; My goal was to get my Ph.D. in CS by age 50. U'r never too old/young"

  • @filler: "Paying attention to people and understanding them is key to truly innovating." @RisingLinda

  • @moshjeier: Self deprecating humor should be a requirement for public speakers

  • @johndesser: Gang of Four pattern book forever changed the way we develop software. @RisingLinda

  • .@RisingLinda: "Myth #1: People are rational

  • .@RisingLinda: "If you want to chg people's minds on Mon, what would u do? Make PowerPoint or decision tree? How'd that work?

  • .@RisingLinda: "Patterns: your role is Evangelist (not a fanatic); the literature show that 'evangelist' is a very good term

  • .@RisingLinda: "Patterns: create short-term goals: build on successes, learn from failures: test the waters, time for reflection, small successes, step by step

  • .@RisingLinda: "Patterns: create short-term goals: test the waters, time for reflection, small successes, step by step

  • .@RisingLinda: "I have a talk that I like, called 'Is Agile A Placebo?"

  • @SoberBuildEng: "Now here's the really bad news from the cognitive scientists: none–zero–of our decisions are rational."-@RisingLinda

  • @filler: "Most people sit there listening to your Powerpoint and your cost benefit analysis ... and they ... don't ... care." @RisingLinda

  • .@RisingLinda: "Why is Evangelist the correct term? Because you don't have any proof, but you have passion & belief"

  • .@RisingLinda: "

  • @filler: "Evangelist is a proper term. How many of us are 'doing agile' because of scientific, double-blind studies? None." @RisingLinda

  • .@RisingLinda: "Evangelists start w/their beliefs; iterate, trying to be scientific, learn, repeat

  • @filler: "We've talked a lot about experiments & failure today. The only way you're going to keep going on is with passion." @RisingLinda

  • .@RisingLinda: "Myth #2: Good always triumphs over evil (should be enuf, right?); (Just World Fallacy cognitive bias)

  • .@RisingLinda: "

  • @moshjeier: .@RisingLinda is so spot on If you aren't passionate in this industry you will either fail or stagnate. Passion is a requirement

  • @davemangot: If you want to change an organization, you have to give them something to believe in, not convince them @risinglinda

  • .@RisingLinda: "Pattern #1: Do Food: an under-appreciated pattern; studies show feeding voter vote more positively

  • .@RisingLinda: "You don't need good ideas [to chg their mind]; you just need good cookies" (haha. wait. uh, oh.)

  • @filler: TIL: a 71-year old, diminutive computer scientist can blow peoples minds with genuine, real talk. Thanks, @RisingLinda

  • .@RisingLinda: "Food connects w/some hardwiring in our head; A friend is someone I break bread with; even for bad ideas

  • .@RisingLinda: "Myth #3: If I just had enough power, I could make ppl change; (I'd fire people in my way. Or kill them)"

  • OMG. @RisingLinda: "It turns out, killing people doesn't change things; it just gets COMPLIANCE." (HILARIOUS) @jdeluccia

  • .@RisingLinda: "Killing people doesn't get you hearts and minds; we don't want compliance;

  • .@RisingLinda: "Pattern: Personal Touch: ie, 'what's in it for me?' to mitigate 'why should I help u?" (more selling patterns)

  • @filler: "What we want is real change, passion, commitment. You don't get that with an edict. You want hearts, not lines." @RisingLinda

  • @SoberBuildEng: "If people won't do agile & continuous delivery, you can fire them. And then kill them" #LOL #notreally Wisdom from @RisingLinda

  • .@RisingLinda: "The problem is the adoption curve; Innovators & Early Adopters is only 15.5%; too small!"

  • @BMC_DevOps: RT @moshjeier: Different people accept new ideas differently @RisingLinda #flowconsf http://t.co/r0aE9H07Pf

  • .@RisingLinda: "

  • @filler: "Mgmt wants to identify these people. These are not people. These are roles. People take turns playing these roles." @RisingLinda

  • .@RisingLinda: "Your job as Evangelist is move people through Adoption Curve; I'm a big believer in ppls ability to chg"

  • .@RisingLinda: "

  • @z3ndrag0n: RT @testobsessed: Myth #3: if I just had enough power I could make people change. Compliance is not change. @RisingLinda #flowconsf http://…

  • .@RisingLinda: "Myth #4: Skeptics, cynics, resistors; THOSE people, well, they must be BAD or STUPID or BOTH!! Ignore them!!

  • .@RisingLinda: "Pattern: Fear Less: use resistance to your advantage. Listen, really listen, & learn all you can"

  • @stack72: 747s are coded with pascal - think about that the next time you fly on one ;) @RisingLinda

  • .@RisingLinda: "They raise an objection, and then I parry; the goal is for one of us to say 'I win, you lose'"

  • @filler: "I walk into a room and I have three strikes against me: I'm old, I'm a woman, and I'm an american. Strike three." @RisingLinda

  • .@RisingLinda: "I listened him into agreement"

  • .@RisingLinda: "There are 48 patterns in Fearless Change; I've covered 3. Go read the book"

  • .@RisingLinda: "Q: how do u motivate yourself on daily basis in face of constant resistance?" "A: stop labeling, help them" (they're asking me for help)

  • .@RisingLinda: "

  • @filler: “Champion skeptics can be valuable. They’re in your org. Get them to help. And take turns wearing that black hat.” @RisingLinda

  • .@RisingLinda: "In science, sometimes you just need to wait for old order to all die. I'm sorry." Haha. This is well known.

  • .@RisingLinda: "Can we measure code quality? We seem to know what beauty is. But..."

  • .@RisingLinda on sudden shifts in history: Nature Of Scientific Revolution

  • .@RisingLinda: "On Cultural Revolution: Mao was huge believer in constant experiments, 1 village at a time" (yikes)

  • @filler: “When people think about org change, they think of large steps. Deceptive. Don’t ignore the small, iterative steps.” @RisingLinda

Panel discussion

Misc

  • @fcnewtech: RT @DevToolSuperFan: Ok universe, you win. I will read The Phoenix Project.

  • RT @moshjeier: Your environment should treat people as valued.

  • RT @cbgrasshopper: Teams, like services, should have well-defined interfaces to the rest of the company. @randyshoup

  • RT @stack72: Too True! RT @moshjeier: "We don't have time to do it right" WRONG, You don't have time to do it twice

  • RT @chemphill: amazon 2 pizza teams: keep teams small enough to be fed by 2 pizzas #flowconsf 1 reason I love working @carbonfive

  • RT @stack72: "Virtuous cycle of quality: Testing -> solid foundation -> Confidence -> Faster and better”

  • RT @cbgrasshopper: Developers: don't report bugs; fix them and submit a pull request / patch WITH TESTS. #flowconsf @randyshoup

  • RT @cbgrasshopper: Do less parallel development; serialize features so you can focus and do them well. @randyshoup

  • RT @adr0sen: It's estimated ~200 billions line of COBOL code runs in production systems around the world today. Wow

  • RT @eshamow: RT @moshjeier: @eshamow Mainframe developers are in huge demand due to the large # of retiring employees

  • RT @filler: “If ops job is to watch dashboards, you’re doing it wrong. Given enough hours, their brains will rot.” @royrapoport

  • RT @cbgrasshopper: RT @eshamow: Incredibly impressed by the talks at #flowconsf thus far. It's about systems thinking in the trenches.

  • RT @gwestr: @j_simone holding court at #flowconsf @heroku @SalesforceEng http://t.co/B6yK12VyFt

  • RT @ryantomlinson: RT @stack72: Netflix have a no jerk policy when hiring

  • RT @moshjeier: Tricky parts: Version config, planning db migration (schema, etc), rollback deployments, canary deployments

  • RT @eshamow/@testobsessed quoting 5th Discipline Long feedback cycles are like adjusting a slow hotel shower temperature dial
    @RealGeneKim: RT @gwestr: @j_simone holding court at #flowconsf @heroku @SalesforceEng http://t.co/18K1cvUsE8

  • @medellre: RT @moshjeier: To start bringing the mainframe world into the CD world, invest in test engineering, most bang for the buck

  • RT @medellre: RT @moshjeier: To start bringing the mainframe world into CD world, invest in test engr, most bang for the buck

  • RT @thinknow: Why r feedback cycles important? Need to quickly see if intentions match implementation @testobsessed @pivotallabs

  • RT @SoberBuildEng: .@testobsessed: If ur agile process has "stabilization sprint," u're effectively fudging the meaning of done.

  • RT @moshjeier: Feature flagging has to live in the db/api/service at application layer, can't be done via config

  • RT @dominicad: #flowconsf has the distinction of 1st tech conference where I could run into ppl I needed to talk to in the ladies room.

  • RT @stack72: Session for all parts of the spectrum at #flowconsf - should have been called systems thinking conf :)

  • RT @moshjeier: RT @pkdotnet: "The Dev environment is the production of developers." Give it the effort it deserves. @sasha_d
    RT @filler: "Our 5 goals:

Make it simple.
Exhibit craftsmanship.
Inspire delight.
Deliver uniq value.
Focus on human goals

flowconsf

  • @RealGeneKim: .@ccourage: "Credible: how do we show design impact? here's what consoles looked like: 20y old" #flowconsf http://t.co/EkgLKTgTPW
  • RT @stack72: RT @moshjeier: Get ppl out of cubicles & board rooms. Get people out into the open, make the collaboration visible.
    Haha. RT @JDeLuccia: 6 servers deployed; 4 w/ wrong vers & 900 "issues" http://t.co/3Ux9dabr4H http://t.co/gE3dvdoGiS
    RT @chemphill: citrix circa 2010: we have a design team? great, I need an icon @ccourage

  • @dominicad: Loving #flowconsf Many thanks to organizers for creating awesome! @thinknow @testobsessed @johndesser @jezhumble @RealGeneKim @geeters

  • @dominicad: Loving #flowconsf Many thanks to organizers for creating awesome! @thinknow @testobsessed @johndesser @jezhumble @RealGeneKim @geeters

  • RT @dominicad: Loving #flowconsf Thx organizers for creating awesome! @thinknow @testobsessed @johndesser @jezhumble @RealGeneKim @geeters
    RT @mblongii: The most common form of entropy in an organization is apathy about the mission.
    RT @filler: Regarding detractors and pessimists, "You will only be as good as the team you have around you." @ccourage
    Haha. Awesome!!! RT @BMC_DevOps: #flowconsf @realgenekim trying @adrianco google glass http://t.co/Fs765MhLyc
    RT @filler: "Now u have some new ideas & u're wondering 'What can I do on Mon morning?' Perfect. Thats why I'm here" @RisingLinda
    RT @SoberBuildEng: "Now here's really bad news from cognitive scientists: none–zero–of our decisions are rational" @RisingLinda
    RT @filler: "Most people sit there listening to ur Powerpoint & your cost benefit analysis... they.. don't.. care." @RisingLinda
    RT @filler:"Evangelist is proper term. How many of us r 'doing agile' b/c of scientific, double-blind studies? None" @RisingLinda
    Moore Adoption Curve RT @moshjeier: Different people accept new ideas differently @RisingLinda #flowconsf http://t.co/r0aE9H07Pf
    RT @filler: "I walk into a room & I have 3 strikes against me: I'm old, I'm a woman, and I'm an american. Strike 3" @RisingLinda