Introduction to Prometheus: Concepts, Deployment, and Integration with Grafana

Prometheus is an open‑source service monitoring system and time‑series database, characterized by three key aspects: it is free and community‑driven, it focuses on monitoring servers, databases, and applications, and it stores immutable data points ordered by time.

Compared with older solutions like Zabbix, Prometheus excels in container monitoring and is recommended for new monitoring selections, especially for backend developers who need basic monitoring knowledge.

The main concepts include:

Prometheus Server : the core component that scrapes, stores, and queries monitoring data.

Exporter : exposes metrics via an HTTP endpoint that the server pulls; many official exporters exist (e.g., MySQL).

PushGateway : allows push‑mode metrics to be collected and then served to the server.

Service Discovery : automatically discovers services in micro‑service environments, avoiding manual configuration changes.

The "three musketeers": Prometheus, Grafana, and Alertmanager work together for data collection, visualization, and alerting.

Deployment steps (using Docker):

docker run -d -p 9090:9090 -v /Users/yinjihuan/Downloads/prometheus/prometheus.yml:/etc/Prometheus/Prometheus.yml prom/prometheus

Sample prometheus.yml configuration:

global:
  scrape_interval: 15s
  external_labels:
    monitor: 'codelab-monitor'
scrape_configs:
  - job_name: 'prometheus'
    scrape_interval: 5s
    static_configs:
      - targets: ['localhost:9090']
  - job_name: beta_node
    static_configs:
      - targets: ['10.111.83.8:9100']

Deploy Grafana:

docker run -d -p 3000:3000 --name=grafana -v /Users/yinjihuan/Downloads/Grafana-storage:/var/lib/grafana grafana/grafana

Install node_exporter from the official Prometheus site, start it (default port 9100), and ensure its IP matches the target defined in the Prometheus configuration.

In Grafana, add a data source pointing to the Prometheus server (e.g., http://localhost:9090 ) and import ready‑made dashboards for node_exporter metrics to visualize machine‑level monitoring.

The author plans to share further experiences and troubleshooting notes after deeper exploration.