Observability: Metrics, Logs, Traces
20 min
Day 101: Grafana: RED/USE dashboards
Two proven dashboard frameworks
Rather than dashboards growing organically into an unreadable wall of graphs, two frameworks give a repeatable structure for what to actually show.
- RED (for services/requests) — Rate (requests/sec), Errors (error rate), Duration (latency, usually p50/p95/p99)
- USE (for resources) — Utilization (% busy), Saturation (how much queued/waiting work), Errors
Why two frameworks, not one
RED answers 'is this service healthy from a user's perspective?' — it's about requests. USE answers 'is this resource (CPU, disk, connection pool) about to become a bottleneck?' — it's about capacity. A service can look fine on RED while a USE dashboard shows its database connection pool is about to saturate.
Key terms
- RED method
- Rate, Errors, Duration — a dashboard framework for services.
- USE method
- Utilization, Saturation, Errors — a dashboard framework for resources.
Your API's RED dashboard shows healthy request rate, error rate, and latency, but users report intermittent timeouts. What might a USE dashboard reveal that RED wouldn't?
