We were excited to participate in the sold-out SREcon EMEA 2022 in Amsterdam with our team! We enjoyed meeting & conversing with many interesting people in the SRE community from all over the world. We were happy to have one of our own Bo Hou talk about “A better way to manage Command Line Tools” and introduce his team’s newly released open source project command-launcher.
Around 400 participants shared moments in the conference talks, breaks & social sessions at Amsterdam’s Old Stock Exchange (Beurs van Berlage), a perfectly suited venue. The organization of the event was well executed, and our thanks goes to the staff, volunteers, and organizers. We can’t wait for next year in Dublin! Here is an insight into our highlights!
Making the Impossible Impossible: Improving Reliability by Preventing Classes of Problems
Harold’s favourite presentation was a meta conversation on methodologies for solving problems which exist in the world. The presenter, Chris Sinjakli, uses 3 case studies: State Machines, Memory Safety and DB Migrations to demonstrate his point. While none of the topics are related inherently he investigates the rational for creating certain solutions and how making the good path the only path can reduce headaches. For the purposes of summary I’ll only talk about the one which relates to his title “Making the Impossible Impossible”, State Machines something that as developers and SRE we deal with daily, some times without thinking about it. His thesis for this example was that using or writing a tool which cleanly understands how a state machine should function to have impossible states or specific transitions between states can save valuable time and resources, lowering overhead and perhaps even the complexity of the API and internal logic. I would fully encourage anyone to who finds this interesting to seek out the video once it’s posted in November.
Over Nine Billion Dollars of SRE Lessons — the James Webb Space Telescope
One of Grégoire’s highlights was a talk was given by Robert Barron working at IBM. He narrated how functional requirements of James Webb telescope translated into technical constraints that are very similar to the kind of qualities we SRE are looking for on our systems. Except that, this telescope took 20 years and 9 billions dollars to build so “failure is not an option”. In his words: we can’t aim for less than 100% reliability and yet we have to work with 10 completely new technologies that were not invented when project started. I think we all would have liked more anecdotes. It was nonetheless a great talk for us engineers, especially for those passionate about space exploration.
Keywords: Absolute reliability, repairability and redundancy!
Caching Entire Systems without Invalidation
Gregoire had a second highlight though — A presentation given by Peter Sperl from Bloomberg. A bold and though provoking claim we should design system to not require cache invalidation, which is a notoriously hard problem. Through simple examples Peter took us through his proposal to split system between pure components without side-effect taking all input as parameters and the rest of the impure world. The former have properties so desirable (cacheability, reliability, scalability) it’s worth working hard to put them in place. A must-watch for all SREs building caching layers, and probably for other engineers as well!
Diamonds with Flaws: Examining the Pressures, Realities, and Future of Site Reliability Engineering
This keynote given by Alex Hidalgo from nobl9 (and former Google SRE) was Hervé’s favourite. In this talk, Alex wanted to comfort those that have not the perfect production environment. To quote him “If you never missed a plane, you are spending too much time in the airport.“
For all issues you can have with SRE, someone somewhere has confidently stated that they have solved the problem and explained it through talks, white papers, tweets, best practices, a new technology… It can be dangerous because actually they solved their issue and not especially yours. You don’t deploy 10 times per day or don’t have the best monitoring system like cool people on stage. And it’s fine. We are not all GAFA. And actually, gafa developers are less than 1% of total developers!
So what’s the answer? Here are some Alex tips: Be meaningful in your decisions and choices; Adopt the right processes and technologies for you, fight against the Ivory tower and talk with the product manager, and developers of your organization and keep in mind the best issue to solve is: get the data, analyze it, use it to take the right decision.
The backstage of the event
SRE Con dazzled us with its flawless organization. A magnificent place with the height of luxury: daylight in the exhibition space. Not to mention the giant screen for broadcasting messages, the different lighting ambiances depending on the moment…. A real happiness for the people spending their days on the partner booths 🙂
And it was the conference of all the firsts:
- Our neighboring booth Rootly didn’t get its back wall and to make up for this lack they hired a local artist to paint their white background. In 10 years of event management for Daphnée, this is the first time she had seen something like this! Kudos!
- For the breaks, we were lucky enough to be able to enjoy huge chocolate bars that were broken into squares in front of our eyes. An original and local animation that pleased all the participants
And if you’re looking for something that SRE Con participants like, she recommends Raspberry PI 3 or legos — they’re a huge hit!
We are extending a THANK YOU to the people who made the experience & organization of this years event flawless. See you next year SRE Con EMEA team!
Meet us next year in Dublin or, even better, represent Criteo at the stand by joining the team!
Careers at Criteo | Criteo jobs
Find opportunities everywhere. Choose your next challenge.
careers.criteo.com
