Introduction

A monitoring and alerting system used to keep an eye on applications and infrastructure.

• Time-series database
  • CPU usage
  • Memory usage
  • Request count
  • Request latency
  • Error rates
• Metrics collection system
• Query engine
• Alerting trigger

PULLS model, workflow:

Prerequisite: expose prometheus at application.yml

1. Applications expose metrics at an HTTP endpoint (usually /metrics)
2. Prometheus scrapes (pulls) those metrics at regular intervals
3. Data is stored in its internal time-series database
4. Query the data using PromQL
5. Alerts fire if conditions are met

Where Prometheus Data Is Stored

• Stored on local disk, inside:
  • WAL (write-ahead log)
  • TSDB blocks

// Default: /data

❌ Single-node storage