RED 및 USE 방법

With endless metrics available, teams drown in dashboards. RED and USE are proven recipes that tell you which few signals actually matter.

RED targets request-driven services and tracks three signals: Rate, Errors, and Duration.

• Rate: requests per second
• Errors: failed requests per second
• Duration: latency distribution

For every service, chart these three. Together they answer how busy it is, whether it is failing, and whether it is slow.

RED reflects what users feel: are their requests served, served correctly, and served quickly. It maps directly to user experience.

USE targets resources and tracks Utilization, Saturation, and Errors for each component like CPU, memory, disk, and network.

For each resource:

• Utilization: percent of time busy
• Saturation: queued work waiting
• Errors: error events

Walk the resource list and apply USE to each, finding bottlenecks systematically rather than guessing.

High utilization can be fine; saturation, the backlog of waiting work, is the early warning that a resource is becoming a bottleneck.

They are complementary.

• RED: service health from the user view
• USE: resource health from the system view

RED tells you something is wrong; USE often tells you why.

A latency spike (RED Duration) might trace to disk saturation (USE). Using both methods links symptom to cause.

Design service dashboards around RED and host dashboards around USE. This keeps panels meaningful instead of a wall of noise.

You learned two focusing methodologies: RED (Rate, Errors, Duration) for request-driven services from the user perspective, and USE (Utilization, Saturation, Errors) for resources from the system perspective. RED reveals symptoms, USE reveals causes, and together they keep dashboards meaningful.