The Surge Conference, organized by OmniTI, gathers the best and brightest in web operations on the (US) east coast to share knowledge and experiences. Chef was a proud sponsor again this year and we provided Habitat training prior to the conference on workshop day.
About 40 attendees gathered for our workshop, Introduction to Application Automation with Habitat (slides), led by Nathen Harvey. Nathen showed how Habitat lets teams focus on the automation around an application without having to make decisions about infrastructure early on. Attendees got hands-on experience packaging and deploying a simple Node JS web application to an AWS host using only Habitat and running within a Docker container. Next, Nathen did a demonstration of clustering Redis with leader/follower election management built in.
Adam Leff followed up with a demonstration of load balancer, front-end, and back-end processes using Habitat’s service groups and discovery to adapt their own configurations as processes were added and removed from the system.
Attendees asked deep questions, showing that they were understanding Habitat concepts and picturing its use in application deployment.
Adam Steltzner, NASA engineer and author of The Right Kind of Crazy kicked off the conference with the high-energy story of landing Curiosity on Mars. He shared many keys to the success of what he likened to interplanetary golf:
It was an excellent and inspiring talk. (I’m buying the book.)
Chef’s Nicole Forsgren spoke on “What We’ve Learned From Four Years Sciencing the Crap Out of DevOps.” Nicole shared many findings on how technology, processes and culture can improve outcomes. She provided advice on how to measure squishy things like team happiness and culture with science-based survey techniques.
Technology is not a cost-center. Tooling for automation and testing pays dividends in being able to deliver change more quickly, whether that change is new features, fixes for bugs, or responses to outage incidents.
Also, data from Google—who has been studying managers for years—concluded that the number one thing a manager does is provide psychological safety for their team. (I’m sensing a theme.)
The conference was closed by Circonus’ Theo Schlossnagle. A point that stood out to me was using inverse quantiles instead of 99th percentiles. This lets you find what parts of your system may be experiencing truly epic badness instead of masking away the badness under the report that “well, 99% is alright.”
At Surge, practitioners are encouraged to share war stories and battle scars in a safe place. The goal is to collectively lift the practice of scaling and operating systems. The infamous “hallway track” is always excellent. The ubiquity of bacon options during meals is also appreciated. We’re looking forward to Surge 2017!