What are the RED and USE methods for monitoring?
Learn the RED (Rate, Errors, Duration) and USE (Utilization, Saturation, Errors) monitoring methods, when to use each, and how to instrument them in Prometheus.
Expected Interview Answer
RED and USE are two complementary monitoring methodologies: RED (Rate, Errors, Duration) focuses on request-driven services, while USE (Utilization, Saturation, Errors) focuses on resources like CPU, memory, and disks.
The RED method, coined by Tom Wilkie, tells you how your services behave from the caller's perspective: how many requests per second, how many are failing, and how long they take. The USE method, coined by Brendan Gregg, tells you the health of every resource: how busy it is (utilization), how much work is queued (saturation), and its error count. RED is best for microservices and endpoints; USE is best for infrastructure. Used together they cover both the service and the machine it runs on.
- RED gives a consistent per-service dashboard template
- USE catches resource exhaustion before it cascades
- Together they cover both application and infrastructure
- Both are simple, memorable checklists
- They guide which metrics to actually collect
AI Mentor Explanation
RED is like judging a bowler by deliveries per over, wides and no-balls conceded, and time taken between balls. USE is like checking the pitch and player condition: how worn the surface is, how fatigued the bowler is, and injuries reported. One measures the service delivered, the other the state of the resources producing it.
Step-by-Step Explanation
Step 1
Apply RED to services
For each request-serving component measure Rate, Errors, and Duration.
Step 2
Instrument request metrics
Expose a request counter, an error counter, and a latency histogram in Prometheus.
Step 3
Apply USE to resources
For each resource measure Utilization, Saturation, and Errors.
Step 4
Instrument resource metrics
Use node_exporter and cAdvisor for CPU, memory, disk, and network signals.
Step 5
Combine the views
RED dashboards for services, USE dashboards for infrastructure, correlated on incidents.
What Interviewer Expects
- RED = Rate, Errors, Duration for services
- USE = Utilization, Saturation, Errors for resources
- Knowing who coined each (Wilkie, Gregg)
- When to apply each method
- How to instrument them in Prometheus
Common Mistakes
- Mixing up what each letter stands for
- Applying USE to request-based services
- Ignoring saturation and only watching utilization
- Treating them as competing rather than complementary
Best Answer (HR Friendly)
“RED and USE are simple checklists for what to monitor. RED (Rate, Errors, Duration) watches how your services handle requests, and USE (Utilization, Saturation, Errors) watches how healthy your machines and resources are. Using both gives you a full picture.”
Code Example
# Rate: requests per second
sum(rate(http_requests_total[5m]))
# Errors: error ratio
sum(rate(http_requests_total{status=~"5.."}[5m]))
/ sum(rate(http_requests_total[5m]))
# Duration: 95th percentile latency
histogram_quantile(0.95, sum(rate(http_request_duration_seconds_bucket[5m])) by (le))Follow-up Questions
- Why is saturation harder to measure than utilization?
- How would you build a RED dashboard for a microservice?
- Which exporters give you USE signals?
- How does the four golden signals framework relate to RED?
- When would RED alone be insufficient?
MCQ Practice
1. What does the D in RED stand for?
RED stands for Rate, Errors, and Duration, where Duration is request latency.
2. The USE method is best applied to?
USE (Utilization, Saturation, Errors) is designed for resources such as CPU, memory, and disks.
3. Which method was coined by Brendan Gregg?
USE was introduced by Brendan Gregg; RED was popularized by Tom Wilkie.
Flash Cards
What is RED? — Rate, Errors, Duration — a method for monitoring request-driven services.
What is USE? — Utilization, Saturation, Errors — a method for monitoring resources.
Who coined USE? — Brendan Gregg, for systems performance analysis.
When use RED vs USE? — RED for services and endpoints; USE for CPU, memory, disk, and network resources.