Dashboard-Templates
SRE-On-Call
- 4 Widgets oben: error rate / p95 latency / RPS / saturation (die USE-Methode)
- Variablen: env=prod (static), service=label_values
- Auto-Refresh: 30s
Capacity Planning
- 7-Tage-Graphen CPU/RAM über die Flotte
- Predict_linear für Memory:
predict_linear(node_memory_used_percent[7d], 86400 * 30)— wo wir in 30 Tagen stehen - Authentication Metrics: neue Nutzer / aktiv / Churn
Business-KPI
- Revenue (eine Custom-Metrik)
- Signups / day
- Funnel: visits → signup → trial → paid
Embed für Kunden
- 1-2 Widgets: success rate + latency
- Public Share mit 30-Tage-TTL