Baoyuantop commented on issue #13188:
URL: https://github.com/apache/apisix/issues/13188#issuecomment-4221678207
The current figures are misleading:
1. The baseline is too lightweight: ngx.say(“benchmark”) incurs virtually no
overhead (~24 μs per request), so any additional logic results in a significant
percentage drop in performance.
2. Amplification effect: Without `proxy_pass` and with no upstream latency,
the entire request takes very little time, causing the framework’s overhead to
be greatly magnified.
3. Real-world scenario: In actual production environments, upstream
response times typically range from 1 to 100 ms. In this case, the ~50 μs
overhead in the access phase accounts for less than 5% of the total time.
Recommendations
1. Perform a comparison using a real upstream server (even if it’s just a
direct Nginx response) via `proxy_pass` and observe the end-to-end latency.
2. Verify that routing is working correctly, and check the number of global
rules and enabled plugins.
3. Refer to
https://docs.api7.ai/apisix/production/performance/performance-testing for
performance testing.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
