Prometheus is a systems monitoring and alerting toolkit originally built at SoundCloud and donated to the CNCF, where it became the second graduated project after Kubernetes. It collects and stores metrics as time-series data, identified by metric name and key/value pairs, with a powerful query language (PromQL) for analysis and alerting.
The pull-based architecture sets Prometheus apart — instead of applications pushing metrics, Prometheus scrapes HTTP endpoints at configured intervals. This makes it easy to monitor any application that exposes a /metrics endpoint, and the ecosystem of exporters covers databases, hardware, messaging systems, and more. Service discovery integrates with Kubernetes, Consul, DNS, and other systems.
Prometheus is designed for reliability and simplicity. Each Prometheus server is standalone with local storage, requiring no distributed storage or external dependencies. This makes it operationally simple but means long-term storage and high availability require additional solutions like Thanos, Cortex, or Grafana Mimir.