The Setup
The standard open-source monitoring stack is Node Exporter (metrics collection), Prometheus (time-series storage), and Grafana (visualization). This architecture mirrors what runs in production Kubernetes environments across APAC enterprises.
The implementation is straightforward. Node Exporter runs on the target server (default port 9100), exposing system metrics in Prometheus format. Prometheus scrapes these metrics at regular intervals (typically 15-60 seconds) via HTTP endpoints defined in prometheus.yml. Grafana connects to Prometheus as a data source and renders pre-built dashboards (community template 1860 is the standard for Node Exporter).
What Matters for Enterprise Teams
Prometheus is CNCF-graduated and handles 10+ billion metrics daily in production environments. Grafana Labs reports 220M+ ARR with 1M+ active users. The pull-based model scales via federation but requires careful network design.
The networking detail matters. Bridged adapters work for lab setups. Production deployments need firewall rules, TLS termination, and authentication (ports 9090, 3000, 9100). The pull model can be problematic in multi-cloud or heavily firewalled environments. Remote write to Grafana Cloud or similar push-based alternatives solves this but introduces vendor dependencies.
Trade-offs Worth Noting
For single-server monitoring, this stack is overkill. Tools like Netdata or traditional agents (Zabbix, Nagios) require less overhead. The value appears when monitoring distributed systems at scale.
The article positions this as "observability" but it's primarily metrics. True observability includes logs and traces (typically via Loki and Tempo). Alert Manager configuration (mentioned but not detailed) is non-trivial and deserves its own implementation guide.
What's Missing
Production readiness requires alert rules, retention policies, and high availability configuration. Prometheus stores data locally by default. Losing that disk means losing your metrics history. The article doesn't cover persistent storage or backup strategies.
Security gets one line ("admin/admin" default credentials). Production Grafana needs SSO integration, role-based access control, and audit logging.
This is a solid learning exercise. The gap between this setup and production-ready monitoring is where the real work starts.