The golden signals are the four things Google SRE says to watch first, and they cover most service incidents.
The four golden signals, from Google's SRE practice, are latency, traffic, errors and saturation. Monitoring these four for a user-facing service captures most of what you need to detect and diagnose problems.
Each answers a different question, and together they give a compact, high-signal view of service health.
| Signal | What it measures | Why it matters |
|---|---|---|
| Latency | How long requests take (and the latency of failures separately) | Rising latency is the earliest sign of trouble users feel |
| Traffic | Demand on the service (requests per second) | Context for the other signals and a load indicator |
| Errors | The rate of failing requests | Direct measure of things going wrong |
| Saturation | How full the service's resources are | Predicts imminent failure as a resource runs out |
TelemetryOps captures the golden signals as first-class inputs, and Sentinel AI uses them to detect, localise and act on service incidents.
Google's Site Reliability Engineering practice popularised latency, traffic, errors and saturation as the four golden signals.
RED (rate, errors, duration) is a request-centric subset for services; USE (utilization, saturation, errors) is resource-centric; the golden signals span both.
Ops Singularity turns open telemetry into autonomous, governed resolution. See it on your own stack.