Explain the RED and USE methodologies and when to use them.

Deep dive

Pick the framework based on what you’re observing:

• RED: APIs, web services, RPCs.
• USE: infrastructure resources, queues, caches.
• Combine with SLOs for user impact.

Examples

RED for HTTP:

Rate = requests/sec
Errors = % non-2xx
Duration = p95/p99 latency

USE for CPU:

Utilization = % CPU
Saturation = run queue length
Errors = throttling or faults

Common pitfalls

• Using RED for non-request resources.
• Ignoring saturation, which predicts future failure.
• Focusing on averages instead of tail latency.

Interview follow-ups

• How would you apply USE to a database?
• What is the fastest signal for a user-facing outage?
• How do RED/USE map to SLOs?