Grafana & friends Munich

Our February meetup is on the 10th floor of the amazing Kraftwerk location by MaibornWolff. It's 350m from U-Bahn U3 Forstenrieder Allee, and there is an underground car park (U2/U3). We can invalidate the parking tickets at the Meetup.

Talk 1) Unraveling Prometheus: Demystifying Counter Metrics and Rate Calculation in Grafana
Ismail Simsek, Senior Software Engineer, Grafana Labs
Abstract: We all have counter-metrics almost everywhere. And Prometheus gives us three functions to see the rate of the increase for counter metrics. For better granularity, you must choose the best time window for your queries. Choosing the right time window is hard. To simplify that, we have built-in variables: $__interval and $__rate_interval. A recently discovered, and fixed, bug in $__rate_interval calculation made me realize how confusing all those terms are. Knowing how those are calculated is not necessary but is the key to better queries. Better queries are the key to better dashboards and alerts.
In this talk, we will explore how Prometheus calculates the rate of increase in different functions, what do we have in Grafana in terms of step and interval, why do we need $__rate_interval, how Grafana calculates the $__rate_interval. By the end of the talk, you'll have a better understanding of rate(), increase(), $__rate_interval and hopefully, you'll have better granularity in your queries.
Bio: Ismail is the tech lead of the Observability Metrics squad and is based in Munich, Germany. His diverse background in development and desire for continuous learning has led to him wearing many different hats. He likes investigating bugs, fixing them, and helping his fellow bug hunters. Besides his professional passions, he enjoys good food with his wife, reading books, discovering worlds in RPG games, and destroying monsters with his daughter on Xbox..

Talk 2) Juju and COS: a Charming solution to Observability
Luca Bello, Software Engineer, Canonical
Abstract: The Canonical Observability Stack (COS) is an open source project which leverages the Juju ecosystem to allow for easy monitoring of multi-substrate systems, incorporating best-in-class software (such as the LGTM stack) to provide smooth day-2 operations and an incredibly simple integration with existing machinery. This presentation aims to showcase COS and the work we've been doing to facilitate its integration with any deployment.
Juju is an open source orchestration engine which allows to seamlessly deploy applications across multiple clouds (both private and public) and substrates (oftentimes Kubernetes and VMs), and to easily integrate them by abstracting the operational effort into code. Applications are wrapped in “charmed operators” – or “charms” for short; the aforementioned abstractions make integrating different charms a straightforward CLI operation. For Observability, this means lowering the entry barrier for everyone, making it easier to monitor their systems and thus leading users towards a safer and more robust architecture.
The Observability team is constantly improving and expanding COS, also providing a lightweight version (COS Lite) for smaller scale operations. The bundle includes state-of-the-art observability tools such as Prometheus (or Mimir) for metrics, Alertmanager for alerts, Loki for logs, Grafana for dashboards, and Tempo for traces; it also features a number of tools to ease integrations, such as Grafana Agent, Prometheus Pushgateway, and several exporters.
Bio: Luca started his career at CERN, monitoring extremely-large-scale distributed systems (including international computing grids) and serving the data-sciency needs of their IT department. Today, he's part of the Observability team at Canonical, working on open-source solutions in the Juju ecosystem to make the wonders of good observability easily available to everyone. Oh, and he likes cute things – he really likes cute things.

Note: We originally announced Tobi Schneck's talk "Running Kubernetes with a multi-tenant Observability Stack in a Manufacturing Line". We have to postpone that to next time. Get well soon Tobi!