Is Region like olap bucket?

[thexiay]

I noticed that region is a partition level in greptime, but I found that region is mainly held by one datanode, and other datanodes hold its backup. This is very similar to the concept of bucket in traditional OLA, so this bothers me. . What I want to clarify here is whether there will be no primary key overlap between regions, that is, will the data of the same primary key appear between two different regions? Will there be automatic expansion and contraction of regions in the future? It doesn’t feel like a good idea to let users manually specify the range of regions.

[v0y4g3r]

I noticed that region is a partition level in greptime, but I found that region is mainly held by one datanode, and other datanodes hold its backup. This is very similar to the concept of bucket in traditional OLA, so this bothers me. . What I want to clarify here is whether there will be no primary key overlap between regions, that is, will the data of the same primary key appear between two different regions? Will there be automatic expansion and contraction of regions in the future? It doesn’t feel like a good idea to let users manually specify the range of regions.

Regions hold disjoint parts of data, so the same primary key will never appear in two different regions. Region merge and split (repartition) is manual in open-source, but automatic repartitioning based on realtime workload and resource requirement can be expected in greptime cloud service in the near future.