Deploy continuously? Yes please! But how?*
How do you go from deploying once a week to once an hour or more? Culture is the biggest barrier. What did New Relic do to make this work?
Continuous deployment is great. Each deploy is small, cheap, and easy to roll back, so it’s very low risk. If I founded a startup today, that’s how I would structure my engineering team. But I’m not founding a startup. I work at one. New Relic used to deploy once a week. Now we deploy once a day. That transition took us 6 months. How are we going to make it to once an hour, or more?
Converting to continuous deployment is difficult, and requires changes to infrastructure, workflow and culture. It’s a state of mind, a set practices, not a feature you can check off. How do you take your team, and build that state of mind? How can you ensure quality while deploying all the time?
This talk will look at the challenges we’ve faced, and examine the solutions we’ve come up with. I’ll talk about the most productive ways to think about workflow and quality so that you can move your team closer to the dream of continuous, low-risk deployment.
This is my first time giving this talk. I have extensive speaking experience -- I used to teach botany and evolutionary ecology in my former career.
You can see an example of me giving a technical talk here: https://www.dropbox.com/s/w6nec70k0esjogl/1030-newrelic2012-versioned-ui-large.mp4
This was the second talk I gave at our annual off-site last year. My intro is a bit flat because I'd spoken 15 minutes previously.
What happens when an evolutionary ecologist decides to work on the interwebs? You get a designer/developer who loves plants and lives in Portland in the sunny and always dry Pacific Northwest with his wife and two kittenz.
That’s a fair crack at describing me. After leaving grad school, I had a stint working at Intel on their site, after which I worked for Reed College for a few years helping them move to using Rails. Now I’m at New Relic, where I do both design and development work.