As a backend developer, I'm interested in the DevOps concept and try to take the opportunity to do 'DevOps-y' work whenever I can. This seemed like a great opportunity to broaden my knowledge of the field.
The conference took place over two days at the Geneva School of Business Administration, in a very modern building with a large auditorium. Since I don't speak French and didn't know anybody else at the start of the event, I decided to live-tweet as many talks as I could. I find this is a fun way to meet people at events. It also forces you to concentrate on what speakers are saying, rather than letting your attention drift away from the topic. (When not live-tweeting, I usually take notes by hand for the same effect.)
On the first day, Thursday 21st February, there was a single track with about half the talks being given in French. Although I couldn't understand everything, I'm a fan of conferences having at least some of the talks in the local language. It seems strange to me that so much tech communication assumes English fluency as a prerequisite.
.@gsaslis and @mikz say that any project needs CI, but OSS projects need it more!— Rae Knowler (@RaeKnowler) February 21, 2019
I enjoyed Yorgos Saslis-Lagoudakis and Michal Cichra's account of horizontally scaling their CI pipeline: it was a good reminder that putting together your own solutions can be less efficient than using a pre-built one, though even then there are ways to optimise. Nat Remez's talk about mistakes and psychological safety was also great. Developers and devops are people, and people will always make mistakes—but the fear of getting things wrong stifles innovation. Nat gave us tips for both management and technical practices to reduce the severity of the consequences of mistakes, allowing people to work without fear.
Looking for a story to link Switzerland with her country, Sweden, the first thing @nat_remez found was the NYSE hanging a Swiss flag to celebrate Swedish company Spotify. Great for a talk on mistakes! pic.twitter.com/iyGI1RkzH9— Rae Knowler (@RaeKnowler) February 21, 2019
The lunch and snacks at the conference were all really delicious. I was surprised to see (local, very tasty) beer served with lunch! In the evening, the conference social had more good food and great live music.
Friday, 22nd February, had two tracks, with workshops in a separate room. That meant I could find an English-language talk in every slot. Unfortunately, I didn't live-tweet one of my favourites, because it was in the smaller workshop room and staring at my phone felt awkward! Still, Ignat Korchagin's talk about porting the Cloudflare software stack to the ARM64 architecture was both informative and amusing.
Ramón Medrano Llamas, a Google site-reliability engineer, gave an enlightening talk about error budgets. Publishing a service-level objective (SLO), such as '99.999% uptime', naturally implies that you have a budget for errors (that 0.001%), and you can manage that budget, for example by picking good times and strategies for updates.
One of the most popular talks of the day came near the end: Jihai Zhou explained how he's bringing awareness of DevSecOps to HSBC's Technology China office. Jihai made a strong case for integrating security concerns with development and operations ones, and showed us his practical strategy for doing so in a large, conservative business.
There are three stages in Jihai’s model of adopting DevSecOps:— Rae Knowler (@RaeKnowler) February 22, 2019
1) Integrate cyber security tools
2) Set up training
3) Establish a cyber security mindset: dedicate resources to it, make sure there’s an expert in every team#DevOpsDays
I had to run off quickly at the end of the day to catch my train home to Zürich, missing the opportunity to socialise more with attendees. Nonetheless, I had a great time at the inaugural DevOpsDays Geneva, and hope to be back next year!
Here are all the live-tweet threads I made over the conference:
- Yorgos Saslis-Lagoudakis, Michal Cichra - Faster, Cheaper, Leaner: Horizontally Scaling a CI Pipeline (thread / unrolled)
- Nataliya Remez - Fail fast, Fail often - make safety a prerequisite for your team (thread / unrolled)
- Ruslan Synytsky - Elastic JVM: Automatic Vertical Scaling of the Java Heap (thread / unrolled)
- PJ Hagerty - Proactive Programming (thread / unrolled)
- Julien Pivotto - Monitoring as an entry point for collaboration (thread / unrolled)
- Arun Kumar Singh - A Journey towards self healing infrastructure (thread / unrolled)
- Ramón Medrano Llamas - Service Levels and Error Budgets (thread / unrolled)
- Jihai Zhou - The Challenge and Solutions to Implement DevSecOps into Large Banks (thread / unrolled)
- Ron Yishai - Bottlenecks in Serverless Applications (thread / unrolled)