The view from London: KubeCon Europe 2025

The most exciting part of KubeCon occurred before the event properly began, at the co-located Observability Day. As the only mobile-focused observability member of the CNCF, we joined to learn from the experts who have been making sense of applications for decades. Embrace’s own Hanson Ho was also there to share his experience building mobile observability at scale as well as past Embrace webinar guests Hazel Weakly and Adriana Villela!

The sessions did not disappoint, as they distilled the technical and human challenges and lessons learned from people who’ve rolled up their sleeves and built observability systems.

In her talk “Operationalizing Observability: The Invisible Parts,” Hazel broke down how teams can take their first painful steps towards adding observability to their technology ecosystems. Her main takeaway? Make it easy for people to get started by connecting some dots for them. For example, show them a case where an API endpoint might be hit, and do the work so they can discover this again in the future. In her words, “get people to the finish line by getting them to the starting line.”

Adriana’s talk, “Putting the Experience in UX: The Importance of Making Data Accessible,” given with Marino Wijay, approached observability as a set of principles we can apply to other domains. What if a job application process was treated as a distributed trace across systems? We’d probably track candidates better. What if flight tracking systems all used the same data specifications to communicate across boundaries? We’d probably avoid incredibly dangerous miscommunication.

These conclusions are worth repeating: Any team can get started on observability with some quick insights and approachable information. And any process that needs to make sense of a system’s data should apply some observability best practices!

Later in the day was Hanson’s talk about his team’s experience building observability into Twitter’s mobile apps. The task in front of him was a mammoth departure from previous discussions around observability, in a number of ways.

For one, the cloud native audience tends to focus on topics about and related to Kubernetes. Hence the name KubeCon! However, Kubernetes doesn’t run on mobile apps, which have their own operating system-specific installation and resourcing processes. So why should the audience care?

Second, Hanson had to turn a number of theoretical points about mobile observability into more-tangible results. One week earlier, he and the present author had delivered a talk at SRECon Americas about the goals that teams should have in building mature, reliable mobile app systems. That talk raised eyebrows for many in attendance, but it was theoretical, whereas the Observability Day topic was a case study based on prior art.

Finally, Hanson’s participation in the OpenTelemetry (OTel) project has focused on the needs of mobile and user-facing developers. There are many ways observability has to work differently on thousands or even millions of users’ devices than in backend systems. Making the goals clear to the current audience, and helping them see the specific goals of implementing OTel properly for mobile, would have to be at the forefront.

The author admits that he would not have set up this daunting scenario had Hanson not knocked it out of the park with his talk “Transmissions from the Tweet Factory: How Observability Transformed Mobile Performance at Twitter.” The story, about the technical implementation of mobile observability at Twitter, demonstrated how getting a sense of the app’s users and their technical experience helped fuel actual growth at the company. The room was full of heads nodding along to the relatable goals of “engaged, issue-free users” while captivated by Hanson’s enthusiasm for the topic.

After the talk, leaders of the OpenTelemetry project from different, unrelated domains rushed the stage to speak about the opportunity to grow mobile. They might be learning about this use case soon enough.

In the cloud native community, observability has become inextricably associated with OpenTelemetry. OTel is growing towards becoming the common standard for data modeling and transmission. To echo what many said, this is the year OTel went mainstream.

At Observability Day, Jason Plumb led a technical overview on encrypting telemetry transmission, while Elena Kovalenko talked through Delivery Hero’s observability centralization around OpenTelemetry to coordinate the company’s many distinct technology and business units.

OpenTelemetry was on the main conference agenda as well, applied to every domain thinkable. Grafana’s Marylia Gutierrez walked the audience through database observability on OTel. A team talked through the effort to codify the signals from continuous integration with OpenTelemetry. In her second of three talks, Adriana talked with Nancy Chauhan about a sustainable approach to data centers by configuring the OTel Collector. Get some rest, Adriana.

Finally, Juraci Paixão Kröhling presented a talk entitled “OTel Sucks (But Also Rocks!),” where he talked through end user interviews about the project and how to improve it. The conversations were clear, sober looks at an ecosystem that many have come to rely on, and now wish to improve.

During the heart of KubeCon, the OpenTelemetry Governance Committee reported on the staggering scale of the project and its progress. At a given time, an average of 500 contributors are working on 80+ repositories related to OTel.

Deep in the content of this session was a note that developer experience could improve around production use. This note was two-fold: that documentation should grow, and that more use-cases should be actively accounted for and kept in mind.

The most important use-case is around users. How can we make sense of the way people are interacting with technology if we’re not focusing on human users, in real-life context, trying to accomplish meaningful things? That’s where our focus will be, so let’s see what happens next.