What is considered high latency for a TiDB cluster?

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

Original topic: tidb集群延迟多少算高

| username: TiDBer_Y2d2kiJh

[TiDB Usage Environment] Production Environment
[TiDB Version]
[Reproduction Path] Understand how much latency in the TiDB cluster starts to affect the business, resource status as shown in the figure.
[Encountered Problem: Problem Phenomenon and Impact]
[Resource Configuration] Enter TiDB Dashboard - Cluster Info - Hosts and take a screenshot of this page
[Attachments: Screenshots/Logs/Monitoring]

| username: xfworld | Original post link

This is not clearly defined, it depends on your own scenario.

| username: tidb菜鸟一只 | Original post link

If the business doesn’t report any issues, then there are no issues.

| username: zhanggame1 | Original post link

As long as there are no business issues, there are no problems. Latency has far less impact than packet loss.

| username: 孤君888 | Original post link

99% of our latency is in the tens of milliseconds. Yours is over 100 milliseconds, which is a bit high. Are you not using SSD disks?

| username: TiDBer_Y2d2kiJh | Original post link

It’s not SSD. The IO of TiKV is around 40% busy during idle business hours, and now around 11 o’clock, the IO is at 80%.

| username: 孤君888 | Original post link

I still recommend using SSD disks.