Note:
This topic has been translated from a Chinese forum by GPT and might contain errors.Original topic: 在进行大批量加载的时候是否需要预先split索引
|
username: dxss-lee
When creating a table in TiDB, you can pre-shard the table using shard_bit. However, the index on the table only has one region. Will this region cause a hotspot when data is being inserted? If we split this region for the index, will it improve the efficiency of data loading?
If the index is increasing or decreasing, it will cause hotspots, and even pre-splitting won’t solve this problem. Hash partitioning might help to some extent. Starting from version 6.0, there is a shard index feature that can shard incrementing indexes of the Int type.
Can the index of GUID type and non-auto-increment be used to reduce hotspots through splitting?
As long as it is not incremental, it should have some effect. Also, after pre-splitting, the new regions are still on the original TiKV. You might need to consider setting the scatter variable to balance them to other TiKVs.
This topic was automatically closed 1 minute after the last reply. No new replies are allowed.
|
username: 
|
username: