Skip to main content
TT
Lab
Learn
Sandbox
Community
Language Learning (in Korean)
Get started
Learn
Learning paths
Courses
Load Testing
Training to look at the tail instead of the average
고급 · Lessons 30 · Lab 10
Start the lab
Curriculum
Load Models and Metrics (RPS, p95, error rate)
The Average Cannot Detect the Tail Getting Worse
reading
Establishing the Baseline
lab
Quiz: Load Models and Metrics
quiz
Test Design and Warm-up
The Load Generator Must Cooperate With Backpressure
reading
The Load Ladder and the Saturation Point
lab
Quiz: Test Design and Warm-up
quiz
Finding the Bottleneck
Rejection Is Protection, Not Failure
reading
Reproducing and Removing the Bottleneck
lab
Quiz: Finding the Bottleneck
quiz
Capacity Planning
Sizing Is About Getting the Order of Magnitude Right
reading
Where does the test's target number come from?
lab
Quiz: Capacity Planning
quiz
Coordinated Omission
The worst moment disappears from your results
reading
The server stalled for two seconds and p99 was 60 milliseconds
lab
Quiz: Coordinated Omission
quiz
When the Load Generator Is the Bottleneck
Load the generator never produced is not the server's limit
reading
The report said 500 requests per second; it was actually 48
lab
Quiz: Limits of the Load Generator
quiz
How to combine percentiles
The average of three p95s is nobody's p95
reading
Averaging the shard p95s came out 108 ms above the real p95
lab
Quiz: combining percentiles
quiz
Arrival process and think time
An average of 100 requests per second does not describe the load
reading
Same average rate, but only one side blew up the queue
lab
Quiz: arrival process
quiz
Where a load test lies to you
Getting a 200 does not mean the work was done
reading
3,000 requests a second — and every one of them was an error page
lab
Quiz: Is this test valid?
quiz
Calling a regression
"8% slower" is missing one number: how big the noise is
reading
The change we reverted as a regression was just noise
lab
Quiz: Calling a regression
quiz
Reference docs
k6 문서
Locust 문서
Google SRE — 부하와 과부하