Significant Performance Differences Between TiDB Instances

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

Original topic: 各Tidb实例之间性能差异大

| username: Leox

There are two TiDB instances in the cluster. I used sysbench for performance testing without setting up load balancing, and only wrote the IPs of the two TiDB instances in the sysbench configuration file.

After inserting the data, I tested it directly with the same configuration file without any extra operations in between.
I found that the requests distributed to the two TiDB instances were roughly 5:1. Further testing each TiDB instance individually, I found that the instance with fewer requests had a much lower score than the other instance. There were no errors in the logs.

How should I further investigate this? It seems that I haven’t seen any statistical charts on Grafana that can separately observe multiple TiDB instances. :no_mouth:

| username: 我是咖啡哥 | Original post link

Are the server configurations of the two TiDB instances the same? Including the network card.
What about benchmarking them separately?

| username: Leox | Original post link

The same, the results of the separate stress tests show a significant difference.

| username: Leox | Original post link

I think I found it, it should be a network issue again, I’ll check it carefully.

| username: system | Original post link

This topic was automatically closed 60 days after the last reply. New replies are no longer allowed.