Prometheus is a metrics monitoring system for answering operational questions about a distributed system: What is the request rate? How many requests are failing? Which service is slow? Are nodes, pods, queues, and databases close to saturation?

It became popular in cloud-native systems because applications expose simple /metrics endpoints, while Prometheus handles service discovery, scraping, local time-series storage, PromQL queries, and alert evaluation.

For interviews, focus on when Prometheus is the right monitoring choice, how the pull model works, how to avoid cardinality problems, which PromQL patterns matter, and where metrics fit beside logs, traces, and long-term storage.

In the Docker/Kubernetes/Prometheus sequence, Prometheus is the observability layer: Docker packages the workload, Kubernetes runs it, and Prometheus measures whether it is healthy.

The diagram shows how a single Prometheus server discovers targets, scrapes their /metrics endpoints on a schedule, stores the samples locally, and feeds both queries and alerts from that same store.