DevOps Days Silicon Valley Day 1

by James Wickett on


DevOps Days Opening

  • Live stream > @BMC_DevOps: The show has begun #devopsdays
  • thanks @pagerduty and @citrix for sponsoring
  • for the stream

Keynote by Jesse Robbins

  • Through work at amazon, @jesserobbins started #velocityconf as a culture hack for web ops
  • sure building systems that scale are fun, but in reality we are doing it for survival
  • in a decade we won't remember that we initially had to make it all up, it will reach the level of a standard
  • customers are incredibly powerful in the equation and we need to focus in their interface to our systems
  • @vmwarecloudops: "The right culture and tools are a requirement for success." Trends we see right now affect what we do for the next decade. #DevOpsDays
  • my internship at IBM was working on the automated postal service machine that @jesserobbins calls the DevOps for mail
  • in our orgs we have misaligned incentives between dev and ops (and security and QA and biz)
  • @vmwarecloudops: "Organizations need to make the change to align their incentives."
  • Opperant conditioning and we need to check it out
  • the survival of the business is about adapting to the new world
  • the resistance level to DevOps in struggling organizations is extraordinarily high
  • the unfolding of circuit city
  • some biz problems you can't agile out of (example of blockbuster)
  • love the reference to conways law, our tools and communication mimic each other
  • fundamental attributes of successful cultures: ** shared mission and incentives: infrastructure as code, apps as services, DevOps/all as teams *** you need to consider your hardware as a commodity, (don't give your servers names) *** servers are like farm animals, it is just harder if you let the kids name them *** Build deep instrumentation into services, push complexity up the stack *** rally around agile, shared metrics, ci, service owners on call,... ** Changing the culture *** any change takes time, changing culture is no exception and you can't do it alone *** exploit compelling events to change culture: downtimes, cloud adoption, devops buzz
  • @vmwarecloudops: RT @kurtmilne: #devops key concept from @jesserobbins "app is customer of infrastructure" #cloudops #DevOpsDays

Continuous Quality: What DevOps Means for QA

  • @jeffsussna
  • agile put the focus on delving value and cloud put the focus on services
  • services create value in use and only when they are in use
  • services require uptime, responsiveness, and transparency
  • as a service consumer providing a service, we need transparency so we can be transparent upstream
  • @jonathan_thorpe: "Failure becomes unavoidable. Organizations are now figuring out how to recover quickly and gracefully from outages"
  • new QA tooling and roles move testers earlier in our design
  • BDD isn't just for functional requirements, you can use for monitoring, security
  • @jeffsussna I like RDD, Rugged Driven Development
  • @jeffsussna you love testing and you gotta check out @gauntlt
  • @jeffsussna is rocking it
  • @jeffsussna: Slides from my Continuous Quality talk
  •, read Boundary Spanning Mirroring

Is your team instrument rated? (or Deploying 125,000 times a day)

  • J Paul Reed @SoberBuildEng
  • @robynbergeron: .@SoberBuildEng "culture as defined for this presentation - incentives + human factors"#devopsdays
  • aviation as a case study for DevOps, devs as pilots, ops as air traffic controllers
  • using other industries help describe what we do in DevOps: aviation, firefighters,...
  • FAA has over 125,000 deployments (take offs) a day
  • @robynbergeron: .@SoberBuildEng "dev & ops - similar to pilots & air traffic control = dynamic circumstances & pipeline for handling traffic"
  • instrument flight is called flying in the system
  • @robynbergeron: .@SoberBuildEng "don't want to be solving problems on-the-fly when you're already in a bad state"#devopsdays
  • @robynbergeron: .@SoberBuildEng "know who is responsible, who the best people for jobs/situations are"#devopsdays
  • @jonathan_thorpe: Holding patterns are work in progress. WIP is bad
  • @akhilp: "Until you define it, you can't optimize it" @SoberBuildEngr #DevOpsDays
  • @davemangot: Aviate, navigate, communicate @soberbuildeng
  • NASA Aviation Safety Reporting System is an anonymizing system that lets pilots report mistakes to the FAA
  • @petecheslock: "We want to know if 4/5 of pilots screw up the approach so we can fix it instead of blaming them. " @SoberBuildEng

Rerun Open Space

  • Learning some rerun for great good with Lee Thompson and Anthony Shortland and Alex Honor
  • resources for rerun: and
  • rerun has built in unit testing ... Cool!
  • rerun also does tab completion
  • if you feel like coding an env value the that means you need a new rerun option
  • really impressed with the rerun conventions and approach
  • check out shocco it will spit out documentation for your rerun app
  • todo, try the rerun demo and write something

Run Deck Open Space

  • Run Deck is an ops automation tool
  • two patterns of use of run deck: GUI or as a service you can call
  • @rundeck shouldn't be thought of as orchestration but more connecting humans to ops processes
  • Ops Procedures as a Service, add the knobs and levers and logging to your ops with @rundeck
  • sold on @rundeck now and setting it up as one of my first tasks when I get back to work
  • @rundeck provides multi-step procedures (one might say orchestration) all with API, logging, access control
  • Process as Code works on top of Infrastructure as Code
  • @rundeck captures the human to tool and tool to human interaction
  • discussion on how you deploy in your environment is interesting in the @rundeck session
  • to get started with @rundeck
  • there is a Anthony Shortland chefconf preso on rundeck

Using Rundeck and Chef to Build DevOps Toolchains - YouTube

Random tidbits

@davedash: Here’s what Pinterest’s deployer looks like: