Prometheus started at SoundCloud in 2012, created by Matt Proud and Julius Volz. It was inspired by Google’s internal monitoring system Borgmon and became the second project to graduate from the Cloud Native Computing Foundation (CNCF) after Kubernetes in 2018.
The system works on a pull-based model — it scrapes metrics from instrumented targets at configured intervals. This is different from push-based systems and turns out to be really practical for dynamic cloud environments. Services expose metrics at an HTTP endpoint, and Prometheus collects them automatically.
PromQL, the built-in query language, is powerful for slicing and dicing time-series data. You can calculate rates, aggregate across labels, and build complex alert conditions. The alerting component, Alertmanager, handles deduplication, grouping, and routing of alerts to various receivers.
Prometheus stores data in a custom time-series database optimized for high write throughput. It’s designed to be reliable even when other parts of your infrastructure are failing — each Prometheus server is standalone with no external dependencies. For long-term storage and horizontal scaling, projects like Thanos and Cortex extend its capabilities. The project has over 55,000 GitHub stars and it’s become the standard monitoring solution for Kubernetes environments. Most major cloud providers now offer Prometheus-compatible managed services.
