Note:
This topic has been translated from a Chinese forum by GPT and might contain errors.
Original topic: 请问TiKv Region的意义
If you forcibly write data that does not belong to a certain region into that region, what problems will arise?
For example, currently there are four regions: the intervals of [start_key, end_key) are [, 1000), [1000, 2000), [2000, 3000), [3000,)
If there is a key 3100 placed into the [2000, 3000) region, what problems will occur?
Is there any way to put it in?
There are methods, for example, I can pre-partition the Regions first. Then, based on the hash value of the key, select a partition ID and write to that partition.
No problem, logically dividing by rowkey range into different regions is just a metaphor.
- From the logical concept of data, the table is divided into different regions according to the rowkey range, and the region is divided into different stores according to the column family.
- From the perspective of physical storage location, the region is the basic unit of data distribution, and the table’s data will be distributed across multiple RegionServers.
I think it’s impossible to implement the operation you mentioned in TiKV, otherwise everything would be messed up.
This topic was automatically closed 60 days after the last reply. New replies are no longer allowed.
|
username: 我家有个臭皮崽
|
username: 
|
username: 
|
username: 
|
username: