Golden Signals
The four measurements that cover most of what matters for a request-driven service: latency, traffic, errors and saturation.
Latency — how long requests take, split between successes and failures, because a fast failure otherwise flatters the average. Traffic — demand, in requests per second or an equivalent. Errors — the rate of explicit failures plus the implicit ones, such as a 200 with the wrong content. Saturation — how full the most constrained resource is, which is the leading indicator the other three lag.
The reason to name them is coverage. A dashboard with these four per service catches the majority of problems, and it takes an afternoon; the alternative is usually a hundred host metrics that nobody reads and no view of whether users are being served.
Measure them at percentiles, not means. An average latency of 200 ms hides a p99 of four seconds, and the p99 is somebody's every request.