Glossary

Prometheus and Grafana

Prometheus + Grafana is the de facto standard for monitoring cloud-native applications. Prometheus collects and stores metrics; Grafana visualises them. Together they provide a complete real-time picture of system health.

Prometheus

Pull model: Prometheus itself "scrapes" metrics from endpoints (/metrics) every N seconds. Its own query language: PromQL.

# Prometheus metric format (text/plain)
http_requests_total{method="GET",status="200"} 1234
http_request_duration_seconds{p99} 0.245

Grafana

Connects to Prometheus (and 50+ other sources) to build dashboards. Supports alerting: if error rate > 1% — sends to Slack/PagerDuty.

Prometheus metric types

Counter — monotonically increases (request count)
Gauge — can decrease (current connections, RAM)
Histogram — distribution of values (latency buckets)
Summary — pre-calculated percentiles