PromCon 2023 took place in Berlin, between 28 and 29 September. The conference gathered 300 participants and was fully focused on the Prometheus monitoring system.
CodiLime could not miss such an event – the company was represented by Maciej Manturewicz, Director of Engineering and Katarzyna Mirońska-Veigas, Business Development Manager.
Let’s dive into Maciej's first-hand impressions and insights from the PromCon 2023 conference.
Conference recap
Providing software engineering services in the monitoring and observability field is one of CodiLime’s specialties, so I went there to be up-to-date with the latest news and implementation use cases of the tool that is industry standard for monitoring applications and services.
It was my first time at the PromCon conference that, as I learned in the opening speech, started in 2016. It was extremely well organized in the great location of the Radialsystem center on the Spree river.
Each talk was worth seeing and I strongly encourage any observability enthusiast to watch the recordings on the Prometheus YT channel 
. However, our time is limited so I will briefly summarize each talk I attended. I hope it will help you prioritize your watch list.
Talk 1: What's New in Prometheus and Its Ecosystem, by Julien Pivotto and Matthias Loibl
A good place to start for those new to Prometheus and veterans. For the first, a short introduction to the Prometheus system and its architecture was provided. For the second, the latest release notes (among others: native histograms, string labels, SNMP exporter, Java client, new receivers for Alert Manager, etc.) were provided in a brief and entertaining way, which made the talk very enjoyable.
Talk 2: Perses: The CNCF candidate for observability visualization, by Augustin Husson and Antoine Thebaud
An overview and demo of the new visualization component for Prometheus. It is still experimental, and hasn’t reached an impressive level yet. Grafana should not be worried. At least, not yet…
Talk 3: Towards making Prometheus OpenTelemetry native, by Goutham Veeramachaneni and Jesus Vasquez
OpenTelemetry is becoming more and more important in the observability world. Native support of OpenTelemetry in Prometheus is what was expected and it is worth seeing what OTel features are now available in Prometheus (e.g. UTF-8 support, so no more underscore trouble in labels).
Talk 4: Zero-code application metrics with eBPF and Prometheus, by Nikola Grcevski and Mario Macias
A showcase of how to utilize Grafana Beyla to instrument and monitor applications without touching your source code with the utilization of eBPF. In my opinion, it is worth seeing how to do this and also what the challenges and limitations are.
Talk 5: Alert analytics, by Monika Singh
A very good talk – it should be watched by anyone who manages SRE teams. Site reliability engineers are often flooded with alarms and their on-pager duty is not easy. During the presentation, Monika showed how to build an alert analytics solution to tune the alerting system configuration so only the important alerts are triggered and not so very often.
Talk 6: Finding useless and resource-hungry Prometheus metrics, by David Calvert
A good talk on how to utilize Grafana’s Mimir tool for metrics monitoring. There is no doubt here – you should monitor everything, even the monitoring solution. :)
Talk 7: Planet scale monitoring: Handling billions of active series with Prometheus and Thanos, by Sebastian Rabenhorst and Mikołaj Liberski
This talk is worth seeing by those that manage vast and highly scalable infrastructures.
Sebastian and Mikołaj showed how they do it on both the backend and frontend sides to give the audience a better insight into the topic.
Talk 8: Using Green Metrics to Monitor your Carbon Footprint, by Ida Fürjesová and Niki Manoledaki
Another must see in my opinion. GreenOps is a pretty new topic but very important. I didn’t know that data centers produce more CO2 than aviation. From this talk, the Prometheus community can learn how to measure the environmental impact of monitored applications with the help of Kepler .
Talk 9: Observing the world's most famous 3D print farm, by Pavel Strobl
An interesting talk on how to connect a farm of 3D printers to a monitoring solution with a little bit of code and Raspberry Pi.
Talk 10: When my Counter Restarted? Addressing Decade-Old Counter Limitation With Created Timestamps, by Arthur Silva Sens and Bartłomiej Płotka
An advanced talk explaining why sometimes you cannot rely on counter metrics and how to make monitoring more accurate with appropriate usage of timestamps. An interesting turtle and rabbit race allegory. :)
Talk 11: Taming the Tsunami: low latency ingestion of push-based metrics in Prometheus, by Filip Petkovski and Pedro Tanaka
Yet another example of how to efficiently monitor big environments with heavy loads. Here presenters focused on optimization of statsD metrics handling.
Talk 12: Prometheus Java Client 1.0.0, by Fabian Stäber
The title says it all. If you are a Java developer and want to instrument your code and analyze with Prometheus you should definitely see this talk (and start using the new Java client).
Talk 13: From Metrics to Profiles and back again, by Frederic Branczyk
For me it is often difficult to find the right border between profiling and monitoring. This talk blurred it even further for me, describing Continuous Profiling. I’m going to watch it again to get the terms straight.
Talk 14: Where's your money going? The Beginners Guide To Measuring Kubernetes Costs, by Erik Sommer
Again the title makes providing a high-level summary difficult. Erik focused on two metrics to build a cloud cost monitoring solution: kube_node_status_capacity and kube_pod_container_resource_requests.
There were five more talks that I’ll catch up with from the recordings .
Final thoughts
I want to share how impressed I was with how friendly and family-like the Prometheus community is. Most of the attendees knew each and were open to inviting newcomers to join them during the conference itself or during the evening meeting at Birgit .
I highly recommend attending the next PromCon for every OSS monitoring enthusiast.
