Prometheus

Prometheus is the de facto standard for open source infrastructure monitoring. It collects metrics from your applications, systems, and services, stores them as time series data, and lets you query, visualize, and alert on that data. Originally inspired by Google's internal Borgmon system, it's now a graduated CNCF project (the second after Kubernetes) with a large community behind it.

The core model is pull-based: Prometheus scrapes metrics from instrumented targets at configurable intervals. This makes it straightforward to see exactly what's being collected and when. Targets can be defined statically or discovered automatically, which is where its Kubernetes integration shines. It continuously discovers pods, services, and nodes as they come and go, so you don't have to manually update your monitoring config every time your infrastructure changes.

Key capabilities:

• PromQL: a flexible query language for slicing and aggregating time series data across any combination of labels
• Multi-dimensional data model: every metric carries key-value label pairs, letting you filter and group by environment, region, service, or any dimension you define
• Built-in alerting: Alertmanager handles deduplication, grouping, silencing, and routing alerts to your notification channels
• Service discovery: native integrations with Kubernetes, Consul, EC2, and other platforms
• Local storage: an efficient on-disk time series database, purpose-built for high-ingestion workloads

For dashboards, Prometheus pairs naturally with Grafana, which can query it directly. Tools like Uptrace and OpenObserve also support Prometheus as a data source, extending what you can do with the metrics you're already collecting.

Prometheus is self-hosted, which means your metrics stay on your infrastructure. The entire project is Apache-licensed and community-governed under the CNCF.