KubeConEU 2022 - Day 2

In article two of three, we'll focus on the second day of KubeCon + CloudNativeCon Europe 2022 in Valencia, Spain!

After an enriching first day, let's recap the keynotes, announcements, and sessions I was able to participate in online. Let's go!

Keynotes & Announcements

On the second day there were no big announcements but several keynotes around SIGs (Special Interest Groups) on Kubernetes, the software supply chain, and sustainability (of technologies, of the community, etc.). There were even several mentions of the DevX (developer experience) that’s dear to our hearts at Cycloid!

Some notes and links from these bulk keynotes:

• To contribute to the CNCF ecosystem, nothing could be simpler. Go here: https://contribute.cncf.io/ (contributions ≠ code!)
• https://openssf.org/ for everything related to security in the open-source world
• “Expect the worst, design it to be the best.” (Emily Fox) or how to prepare for the worst while achieving the best

For once, I was not thrilled by these keynotes, except on the points indicated above. Maybe it's because I was looking forward to participating in the exceptional sessions of this day? Certainly! Below are the top 3 sessions that I enjoyed.

Session 1 - OpenTelemetry: The Vision, Reality, and How to Get Started

Do you know OpenTelemetry? You know that small open-source project that is the second most active project on GitHub, just behind Kubernetes? Dotan Horovits allowed me to open my eyes wide to the interest and functioning of OTel (its nickname).

A summary?

OpenTelemetry is a unified set of vendor-neutral APIs, SDKs, and tools that generate and collect telemetry data to export it to various analysis tools (Prometheus, Jaeger, or other, for example).

OTel offers an API and an SDK per language for many programming languages (Python, Golang, Ruby, PHP, etc.). This allows you to generate and send traces, logs, and metrics consistently from your applications.

The OpenTelemetry Collector then intervenes to propose a vendor-agnostic implementation on how to receive, process and export telemetry data without maintaining several agents/collectors at the same time.

Finally, OTel is not entirely GA (General Availability). The project is separated into several groups, some are GA (for example, the traces), but for the metrics (it will come in a few weeks!) and the logs (later this year), it’s not yet the case.

As you can imagine, the subject is vast and exciting. Perhaps a future article on OpenTelemetry and its use at (and with) Cycloid will see the light of day, but one thing is sure, we haven't finished hearing about it! Excellent introductory talk that I recommend to everyone!

Session 2 - How to Migrate 700 Kubernetes Clusters to Cluster API with Zero Downtime

Mercedes Benz no longer delivers a keynote but still presents a complete and well-designed talk! Tobias Giese and Sean Schneeweiss allowed us to learn more about their use of Kubernetes and how they managed to migrate more than 700 clusters on Cluster API without downtime!

But what is Cluster API at first? Cluster API is a project to provision, upgrade and operate entire Kubernetes clusters directly via an API (or rather via the Kubernetes API)! Thus, the idea is to simplify these operations using a declarative API. 

The talk highlighted that the operation of the 700 clusters (and more than 3500 machines) in the world is done on an OpenStack on-premise infrastructure. With provisioning that consisted of the use of Terraform, custom self-written tools, and Kubernetes operators. What a legacy!

However, the migration, carried out in 3 phases, caused no production downtime and 0 customer impact. How? By following the migration in 3 stages:

1. Migration of the infrastructure
2. Migration of worker nodes
3. Migration of the control plane

This talk is also enriched with advice and lessons learned from these experiences. And not to spoil the surprise, I advise you once again to watch the replay as soon as it is available on the CNCF YouTube channel!

Session 3 - Cloud-Native Chaos Engineering with LitmusChaos

To finish this article, and because I am always fascinated by chaos engineering, I had to write a review of the Litmus Chaos presentation!

In summary, this open-source platform makes it possible to identify weaknesses in infrastructure by carrying out various chaos tests based on the principles of chaos engineering.

I learned there that a whole ecosystem exists around this platform, like the ChaosHub, a marketplace where you can pick up proven experiments. Or the Chaos Workflow since the release of Litmus 2.0 allows complex experiments to be carried out with parallel execution.

It would even be possible to get out of the Kubernetes context and run scenarios on cloud services like Amazon RDS or Lambda, for example.

It's definitely a platform that I will test and - why not - implement it in various ways in our Cycloid community catalog!

And that's it for the second summary of KubeCon + CloudNativeCon EU 2022! As always, let me know what you thought of this year’s KubeCon on Twitter, where I often "live-tweet" events in which I participate! See you soon for the third and last article of the series!

Day 1 of KubeConEU 2022: read →

Day 3 of KubeConEU 2022: read →