If you are a geek and the Open Source world does matter to you, then you probably already know about a big event called the FOSDEM, which takes place every year in Brussels during the last weekend of January. If you don’t know about it though, go check it out here .
FOSDEM, which stands for Free and OpenSource Software Developer European Meeting, was held inside the Université Libre de Bruxelles (ULB) campus and gathered more than 5000 Free and Open Source Software (FOSS) enthusiasts and developers this year, coming from all around the world for showcases, talks and hacking sessions.
At Criteo and in the R&D department more generally, we use a lot of solutions and technologies coming from the FOSS movements. The core values they share are important to us, and we try to give back as much as possible.
We have the chance to be able to assist many conferences across Europe and the world to stay at the edge of trending technologies, participate and share our experience with communities about tools we integrate and use everyday.
On many aspects, FOSDEM is one great conference that can’t be missed and we were many Criteo fellows attending, be it for professional or personal reasons (Belgian beer, anyone ?).
Even if FOSDEM has some quirks in its organization (some rooms were too small compared to some tracks popularity, for example), the schedule was very attractive this year again.
FOSDEM is also about lightning talks and developer rooms to discover FOSS projects, as well as communities. The year’s forum was held in the K building (just like previous years) and it’s a place where you can talk, share, laugh, troll and gather some goodies of course !
# Tech part
Regarding the technical part, one of the most popular tracks was with no doubt the Container and Process Isolation one.
* The introduction to LXD was very interesting. This project, based on LXC, has a good potential despite being less popular than Docker and friends. Nevertheless, for some companies and organizations, it could be an easier solution to integrate and use than the old and venerable OpenVZ as a true alternative to VMs.
* Then there were several talks conducted by Linux kernel gurus who tried to present upcoming features. There were some good presentations, but maybe too precise and deep into the arcane of Linux, definitely not for a rookie audience.
* Kubernetes, the containers orchestration tool driven by Google also had a large coverage with many demos:
It was pretty clear that other tracks tried to gather people in their rooms with sessions containing the words “Docker” or “containers” in their title.
For example, in the Virtualization dev room, we had a quick overview of the Atomic project from RedHat and how containers can work with VMs, a nice inception session 🙂
On the Config Management side, everybody was focused on the place those tools occupy in the container world.
When we saw the “Beyond config management” topic we all assumed that it was exactly what were thinking, so we were all happy to attend to this talk. For a few years now, people tend to use more and more configuration management tools and of the raise of the “what’s next ?” question became logic. All people used to think the next big step is orchestration, but during this conference the speaker demonstrated this notion is wrong and spoke about “modeling”. An interesting talk overall that should definitely have an impact on the way we use to think.
A great surprise in the Config Management devroom came from an introduction of the DNS management at Facebook done by Stephan Gorget. It was a technical and very well structured presentation, introducing all the tools and components used by Facebook to solve their different problems, entirely based on FOSS tools (TinyDNS, Unbound, Zookeeper, BitTorrent, ExaBGP, Sparts, Thrift, Git, LXC, … and a good share of Python Glue). It was clearly the kind of presentation that could inspire us a lot at Criteo.
Among other interesting topics we assisted :
This was the first talk (9am on a Sunday morning, think about it …) of the famous configuration management track and the room was almost full. The idea of this talk was to reuse the configuration management tools to monitor your infrastructure. We do the same at Criteo and it’s really useful as configuration management is more and more present. This concept should be a best practice when your are using infrastructure as a code. One of us sent a feedback to the speaker to improve the infrastructure presented.
The speakers started their presentation by comparing both ways to configure Jenkins jobs (Manually through the GUI and using Jenkins jobs DSL). The infrastructure part was about automation (Chef, Puppet, CFengine…) and the jobs in Jenkins. The demo on the whole delivery pipeline based on Jenkins DSL plugin was overall really interesting. We are already using a similar setup with our Chef based infrastructure in Criteo.
This talk was about Prometheus which we started to use few months ago. There were no surprises for as we are already using this product and talked with some developers of the tool on their IRC channel. At Criteo we love metrics, and we have a lot of shiny dashboards with awesome numbers. But it’s not easy to query and instrument metrics once you have some thousands of them. The Prometheus approach is really interesting in this case, that’s why we started to work with it.
On many aspects, FOSDEM is above all a social event. Many FOSS communities take the opportunity to organize major sessions and meetings to choose the right path for their movement for the next year or so. It was also the occasion for many of us to see some friends and former colleagues who share the same enthusiasm for the FOSS movement. Many thanks to all the volunteers that make it happened.
Site Reliability Engineer, R&D Xavier works on “core” infrastructure components and automation. Our team’s mission is to empower our other teammates to deliver services in an always more reliable and efficient manner.See Dev Lead roles