Severe Degradation in TiKV Write Performance

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

Original topic: TIKV写性能下降严重

| username: TiDBer_TLK8Du2x

[TiDB Usage Environment] Production Environment / Testing / Poc
[TiDB Version]
[Reproduction Path] What operations were performed when the issue occurred
[Encountered Issue: Problem Phenomenon and Impact]
[Resource Configuration] Go to TiDB Dashboard - Cluster Info - Hosts and take a screenshot of this page
[Attachments: Screenshots/Logs/Monitoring]

I just started learning TiKV this month. When using the Go-YCSB benchmarking tool to stress test data, I found that as the amount of written data increases, the insertion OPS decreases.
As shown in the figure below:


Data parameters are as follows:
image
The CPU and memory of the TIKV nodes are 20 cores 64GB and 64 cores 256GB respectively. Monitoring shows that the utilization of CPU and IO is very low. Could the experts please give me some guidance? :face_holding_back_tears: :face_holding_back_tears: :face_holding_back_tears: