Why is the max latency relatively high when using JMeter to perform custom statement stress testing on TiDB?

Note:
This topic has been translated from a Chinese forum by GPT and might contain errors.

Original topic: jmeter自定义语句压测tidb,为什么max latency会比较大?

| username: magic

[TiDB Usage Environment] Testing/
[TiDB Version] v650
[Reproduction Path] JMeter custom statement stress test on TiDB
[Encountered Issue: Problem Phenomenon and Impact]
When using JMeter custom statements to stress test TiDB, why is the max latency relatively high? As shown in the figure;
Could it be due to connection time during initialization? Not very familiar with JMeter, couldn’t find relevant answers.
[Resource Configuration]
[Attachments: Screenshots/Logs/Monitoring]

Stress Test Output

| username: 我是咖啡哥 | Original post link

max is not very meaningful. Look at avg.

| username: xingzhenxiang | Original post link

Shouldn’t we look at the 90%, 95%, and 99% cases?

| username: magic | Original post link

I am mainly concerned about why max is so large.