This year was my first time in Seattle — and what a way to experience the city. Beautiful weather, great coffee, and an engaged tech crowd. GrafanaCON 2025 brought together developers, engineers, and hobbyists to explore what’s possible with open source observability tools.

From the moment I stepped into the venue, the vibe was energetic but relaxed. 30+ talks, hands-on labs, a science fair (with a working wind tunnel, no less), and a bunch of people sharing how they’ve used Grafana to solve problems in interesting, sometimes unexpected ways. Here’s what stood out to me.

Product Announcements

Grafana 12 shipped with a set of useful features that brought quality-of-life improvements — unified drilldowns across telemetry types, new dashboard tabs, better panel logic, and improved performance. Features like GitHub sync for dashboards and SQL Expressions for data joins give more control to users managing complex environments.

Logs Drilldown in Loki was another highlight. It’s a more visual, intuitive way to explore logs, removing the need for LogQL in many cases. Dropbox’s use case was particularly interesting — they’re using it to manage log data across a massive infrastructure footprint.

The preview of Mimir 3.0 focused on efficiency at scale — better query performance, more reliable storage, and cost optimizations for Prometheus users managing long-term data. It’s not flashy, but it’s the kind of thing teams dealing with scale will appreciate.

Grafana Alloy continues to evolve as a flexible OpenTelemetry distribution. Seeing it hit over 525K active instances was surprising, but the bigger shift is how tightly it’s being integrated with broader OTel efforts. The announcement that Grafana Beyla has been donated to CNCF as part of OpenTelemetry shows a willingness to hand things over to the community.

On the testing side, k6 hit its 1.0 release. Native TypeScript support, semantic versioning, and easier extensibility make it more usable for teams writing custom tests. It’s a solid step forward if you're already using k6.

The new Drilldown apps for metrics and profiling also stood out. They simplify exploring telemetry data — clicking through high-cardinality metrics or CPU/memory usage without needing PromQL or interpreting flame graphs. These tools don’t reinvent the wheel but do make everyday tasks faster.

Real-World Use Cases That Stood Out

One of the most practical dashboards I saw came from Electronic Arts — over 200 error metrics displayed in a clear, color-coded format, giving their dev and QA teams a common language to monitor app health. Nothing fancy, just very well-executed.

Dropbox’s approach to running Loki at petabyte scale is a solid example of how far log aggregation tools have come. Their story didn’t oversell anything — they shared honest takeaways from scaling and optimizing under real pressure.

Microsoft shared their internal migration to Grafana for executive dashboards and incident response workflows. They even open-sourced a custom Azure AD auth plugin, which was a good reminder that even large orgs still need to bend tools to fit.

Glovo’s team talked about using Traces Drilldown to troubleshoot microservices. Filtering through traces visually has apparently saved them time and helped improve reliability — straightforward but valuable.

Other stories came from less expected places. Homer Electric in Alaska is using Grafana to monitor their power grid across remote substations with poor connectivity. And a speaker shared how they monitor their family’s apple orchard using MQTT sensors, solar power, and InfluxDB — a setup that lets them track soil moisture and pump issues from miles away.

Project Bob deserves a mention too — a 14-foot autonomous drone boat streaming telemetry back to shore via satellite and Tailscale, with Grafana as the visualization layer. The project is part science experiment, part open source showcase.

Community and Interactive Experiences

The hands-on labs were well-paced and helpful. I joined the "Grafana as Code" session and walked away with new ideas for automating dashboard deployment. Other labs went into areas like custom data sources, telemetry pipelines with Alloy, and improving dashboard UX.

The Science Fair was more than just a gimmick. A 3D printer monitored live in Grafana, a wind tunnel with pressure sensors, IoT devices tracking air quality — it was a reminder that observability isn’t just for backend services. Someone even had a real-time NCAA dashboard for March Madness, just for fun.

Birds of a Feather sessions gave space for informal, topic-driven chats. I dropped into one on Grafana and OpenTelemetry and found it helpful to hear what others were experimenting with. Other sessions focused on community roles, underrepresented groups in tech, and the nitty-gritty of Tempo, Loki, Mimir, and Pyroscope.

The evening reception at MoPOP was casual and fun — good conversations, interesting exhibits, and a chance to meet people behind the dashboards. Always nice to put faces to GitHub handles.

Wrapping Up

GrafanaCON 2025 packed in a lot. From new product capabilities to niche use cases and hands-on experiences, it balanced updates with community interaction in a way that felt natural. I don’t work at Grafana, but I appreciated seeing the direction the ecosystem is heading and hearing from people solving real problems with it.

Observability might not always be headline-grabbing, but it’s foundational to how we build and run modern systems. And events like this give space to learn, share, and step back from the day-to-day.

Until next time.