add misunderstanding of performannce (#327)

Signed-off-by: Mengxin Liu <[email protected]>
(cherry picked from commit f3eb84e)

Author: oilbeater

docs/advance/performance-tuning.en.md

@@ -9,6 +9,15 @@ so it is recommended to use the latest version to get better default performance
 
 For more on the process and methodology of performance optimization, please watch the video [Kube-OVN 容器性能优化之旅](https://www.bilibili.com/video/BV1zS4y1T73m?share_source=copy_web){: target="_blank" }.
 
+## Common Misunderstandings about Network Performance
+
+Performance tuning and comparison is an exciting topic, and performance data comparisons under different configurations can be fascinating, but all of this may have minimal effect in real application scenarios. Below are some common misconceptions about network performance to help you better judge whether network tuning is needed.
+
+1. The latency introduced by CNI is about tens of nanoseconds per packet. If your application's request processing takes tens of milliseconds, even if CNI has no latency at all, the impact on application latency performance is minimal.
+2. Common network performance tests are packet sending tests. If one network plugin needs 10ns to process each packet while another needs 20ns, there will be a double performance difference, but for real applications with tens of milliseconds processing time, the difference is not significant.
+3. Contrary to first impressions, the bottleneck affecting network performance is usually the CPU. Better and more CPUs bring more significant performance improvements.
+4. If you need extreme network performance, it's best to use network solutions like Macvlan or SR-IOV that adopt ultra-lightweight or hardware virtualization technologies.
+
 ## Benchmarking
 
 > Because the hardware and software environments vary greatly, the performance test data provided here can only be used as a reference,

docs/advance/performance-tuning.md

@@ -7,6 +7,15 @@
 
 更多关于性能优化的过程和方法论，可以观看视频分享：[Kube-OVN 容器性能优化之旅](https://www.bilibili.com/video/BV1zS4y1T73m?share_source=copy_web){: target="_blank" }。
 
+## 网络性能的常见误解
+
+性能调优和对比是令人兴奋的话题，不同配置下的性能数据对比也会让人着迷，但是这一切在实际应用场景下产生的效果可能微乎其微，下面列出一些常见的网络性能的误区，帮你更好的判断是否需要进行网络调优。
+
+1. CNI 带来的延迟大约为每个数据包数十纳秒，如果你的应用本身请求处理在数十毫秒，即使 CNI 完全没有延迟，对应用的延迟性能影响也不大。
+2. 常见的网络性能测试为发包测试，此时如果一个网络插件需要 10ns 处理每个数据包，而另一个需要 20ns，性能会有一倍的差距，但是对于真实数十毫秒的应用来说差距并不明显。
+3. 与第一印象相反，影响网络性能的瓶颈通常在 CPU，更好、更多的 CPU 带来的性能提升幅度会更明显。
+4. 如果需要极限的网络性能，最好使用 Macvlan 或 SR-IOV 这类采用超轻量或硬件虚拟化技术的网络方案。
+
 ## 基准测试
 
 > 由于软硬件环境的差异极大，这里提供的性能测试数据只能作为参考，实际测试结果会和本文档中的结果存在较大差异。