FoundationObservability

What should you measure first for a service? Explain the golden signals (or RED/USE).

What they are really testing: Whether you have a principled starting set (latency, traffic, errors, saturation) rather than measuring everything randomly. Knowing RED vs USE shows depth.

A real interview question

What should you measure first for a service? Explain the golden signals (or RED/USE).

What most people say

drag me

I would measure CPU, memory, and disk usage of the servers.

Those are resource metrics (part of USE) but not what users feel. The golden signals start with latency and errors of user requests; leading with CPU misses the symptoms that actually matter.

The follow-ups they ask next

  • What is the difference between RED and USE?

    RED (Rate, Errors, Duration) is for request-driven services, the user-facing view. USE (Utilization, Saturation, Errors) is for resources like CPU/memory/disk. Use RED for services, USE for the resources behind them.

  • Why measure latency and errors before CPU?

    Latency and errors are symptoms users feel and what SLOs are built on; CPU is a cause you investigate after a symptom appears. High CPU with healthy latency may not matter; high latency always does.

What the interviewer is listening for

  • Names the four golden signals
  • Knows RED for services, USE for resources
  • Starts with user-facing symptoms

What sinks the answer

  • Leads with CPU/memory only
  • No principled starting set
  • Confuses resource metrics with user signals

If you genuinely do not know

Say this instead of freezing. Reasoning out loud from what you do know beats silence every single time, and a good interviewer is listening for exactly that.

Start with [the four golden signals: latency, traffic, errors, saturation]. Use [RED (Rate, Errors, Duration) for request services] and [USE (Utilization, Saturation, Errors) for resources]. Measure [user-facing latency and errors first, they map to experience and SLOs], then [resource saturation to explain them]. CPU is [a cause to investigate, not the first symptom].

Keep going with observability

All 336 cloud engineer questions

Knowing the answer is not the same as recalling it under pressure

Sign in to send the questions you fumble to spaced recall, so they come back right before you would forget them, and learn the concepts behind them with hands-on labs.

Start free