Estimate project
← All posts

KubeCon EU 2024 - a recap of a conference that paid for itself many times over

Catch up on the newest KubeCon takeaways! One of our outstanding developers has prepared a thorough recap you really shouldn't skip.

4 min read
DevOpsConferenceKubernetesKubeConCloud

KubeCon has hit the mainstream, and it is hands down the best conference I have attended so far. Even so, ‘amazing’ does not quite capture how great it was. The people, the talks, the organisation – everything was spot on. And then there is the knowledge – so much of it that my TODO list will keep me busy for the next few months. Put simply, if you can pick just one conference a year, make it KubeCon. Why? As someone attending an event this big for the first time, I want to give a short recap of what I heard, learned, and took away from a few unforgettable days in Paris

KubeCon Paris 2024

👀 Observability

Observability is as vital as breathing. By putting observability in place, you do yourself a favour, because you gain insight into how your system actually runs. The topic dominated KubeCon as well as the Cloud Native Foundation YouTube channel, where every talk from the conference is available. KubeCon opened with dedicated co-located events such as Observability Day, ThanosCon, and CiliumCon. That said, observability at KubeCon went well past the basics. Besides introductory sessions on adding OpenTelemetry to a system, scaling observability systems got a lot of attention. Time moves on, and scaling observability has turned into a challenge for FANG companies and startups alike, particularly in AI and IoT, where huge volumes of data have to be processed and a matching amount of instrumentation data gets published. KubeCon gave me the chance to speak directly with maintainers, authors, and contributors of projects like Prometheus, Thanos, OTEL, and Jaeger, and to hear their take on scaling observability solutions to fit the needs of the project I am working on. It is a huge problem: at peak capacity we hit 28 000 cores on a single cluster running thousands of pods. We run 10+ such clusters and have just one Grafana instance to look across all of them. It is also worth pointing out that OpenTelemetry has begun applying for graduation as a CNCF project. Simply put, this is a sign you should think about moving to OTEL.

On top of that, Fluentbit v3 was officially launched at KubeCon, so the best is still ahead of us.

ArgoCon

ArgoCon took place on day one. I have to say I was stunned, as every talk was full of Argo enthusiasts. ArgoCon covered far more than ArgoCD, which is just one project in the Argo ecosystem. It began with the Argo Workflow project, including how large companies like Bloomberg schedule workflows with Argo Workflow, and ended with Argo Notifications. A single day at ArgoCon is enough to take the mystery out of the Argo system. Since we are a GitOps-first company and use tools from both the Argo and Flux ecosystems, it was well worth hearing insider views on scaling the tools we rely on every day. In the GitOps space, it has always been hard to tell whether a GitOps deployment succeeded, because metrics were missing. This is why Keptn is working with ArgoCD and Flux to make deployments observable, yay 😋

Cilium and eBPF

Before KubeCon, I knew very little about eBPF technology. After going to demos and talks and chatting with people from Isovalent, I now have a rough idea of what it is about.

eBPF technology has been around for a while, evolving from BPF in the 90s into eBPF, which now leads the Cloud Native networking space. eBPF lets you load code into Kernel space on the fly and run logic there, so you can monitor your system at the Kernel level and change how network traffic is routed. Thanks to projects such as bumblebee, the barrier to entry for eBPF technology has dropped – you can give it a go. Cilium, a technology built on eBPF, offers ready-made tools for cluster-to-cluster networking and observability right out of the box. Get ready for the certificate, too. A new Cilium certification has arrived in the Cloud Native space, along with labs.

🔥 Dagger

Pulumi lets you describe infrastructure components in imperative programming languages such as TypeScript, much as you would declare them in Terraform. Dagger lets you write your CI/CD pipeline in programming languages such as TypeScript, Python, or GoLang, leaving behind the classic YAML files used by GitHub Action, Bitbucket pipelines, CircleCI, and others. This makes it easier to run pipelines on your localhost, and it also lets you combine your current pipelines, defined in the GitHub Actions schema, with Dagger. A game changer. I will keep an eye on how the project develops and finds its place among the everyday tools of DevOps teams. And I will certainly try it out myself.

🔥 My personal favourite – Mirrord

Ever heard of Telepresence? If not, you should take a look, but at KubeCon I believe I discovered its successor – Mirrord. It lets us run applications locally while integrating smoothly with the cloud environment, blurring the boundary between local and cloud development environments. Under the hood, Mirrord catches every system call to the filesystem and network socket (by intercepting calls to libc, which most programming languages rely on) and passes them on to the Mirrord agent running in the cloud. I will think about using Mirrord when building a Kubernetes Operator for managing cloud-specific resources.

Wrap-up

KubeCon in Paris has ended, and it repaid every penny many times over. It even inspired me to write my first blog post. It is great to watch the Cloud Native ecosystem flourish, especially in a city as iconic as Paris. Now all eyes are on the next KubeCon, April 1–4 2025, in London. See you there!

Greetings from Paris <3