Load Base Split: Dispelling Doubts About Region Splitting

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

Original topic: Load Base Split 打散region 到疑惑

| username: Raymond

Excuse me, teachers, after using Load Base Split to split a hotspot region, what is the principle behind this splitting? Is it that a new region is split from the hotspot region A, forming region A and region B? At this point, will the sizes of region A and region B be below the specified size for a region, causing a merge of regions again? I’m quite curious about this issue.

| username: xfworld | Original post link

Check the official documentation, there are detailed answers:

| username: ddhe9527 | Original post link

Load Base Split determines Region splitting based on read traffic, with two thresholds: split.qps-threshold and split.byte-threshold. If, within 10 consecutive seconds, the sum of various read requests per second for a certain Region exceeds either the QPS threshold or the traffic threshold, that Region will be split. Regions that have undergone Load Base Split will not be quickly merged. On one hand, PD’s MergeChecker will skip hot Regions, and on the other hand, PD will also judge based on the QPS in the heartbeat information to avoid merging two Regions with very high QPS.