What is The Four Golden Signals?

The four golden signals, popularized by Google's SRE practice, are latency, traffic, errors, and saturation: the core set of metrics that, monitored together, give a high-level picture of a production service's health and user experience.

Start a Free Mock Interview

Free to start · 7-day trial on paid plans

IN DEPTH

In depth.

When you can only watch a few things about a running service, these four tell you the most. Latency is how long requests take (and crucially, separating the latency of successful versus failed requests, since fast errors can hide problems). Traffic is how much demand the service is handling (requests per second, transactions). Errors is the rate of failing requests (explicit failures and wrong-but-200 responses). Saturation is how full the system is, how close key resources (CPU, memory, I/O, connections) are to their limits, which predicts impending trouble.

Together they cover both symptoms users feel (latency, errors) and causes you can act on (traffic, saturation). They are a foundation of observability and of defining SLIs and SLOs: good service level indicators are often built directly on these signals (for example, an SLO on p99 latency and error rate).

For QA, especially shift-right and reliability-focused testing, the golden signals are how you verify behavior in production and during load and resilience tests. Rather than only asserting functional correctness, you watch latency, errors, and saturation under realistic and stress conditions to confirm the system stays healthy, connecting testing to real operational quality.

WHY IT MATTERS

Why interviewers ask about this.

The four golden signals come up in SRE-adjacent, reliability, and observability interviews. Naming latency, traffic, errors, and saturation, and connecting them to SLIs/SLOs and to shift-right and load testing, shows you think about quality in production, not just pass/fail before release.

EXAMPLE

Example scenario.

During a load test, functional checks pass, but watching the golden signals tells the real story: latency climbs steadily, errors spike past a threshold, and CPU saturation hits 95%, revealing the service degrades well below the target throughput. The team fixes the bottleneck before it would have caused a production incident.

TIP

Interview tip.

List the four golden signals, latency, traffic, errors, and saturation, and briefly define each (with the nuance of separating failed-request latency). Connect them to SLIs/SLOs and to verifying health during load, resilience, and shift-right testing, to show you tie testing to real production quality.

FAQ

Frequently asked questions.

What are the four golden signals?

Latency (how long requests take, separating successful from failed), traffic (how much demand the service handles), errors (the rate of failing requests), and saturation (how full key resources are relative to their limits). Together they summarize a service's health and user experience for monitoring.

How do the golden signals relate to SLIs and SLOs?

They are a natural foundation for them. Service level indicators (SLIs) are often built directly on golden signals, such as p99 latency and error rate, and service level objectives (SLOs) set targets on those indicators. Monitoring the golden signals is how you measure whether a service meets its reliability goals.