CODEWITHSUNDEEP
databasehbaseclickhouse
Gomo
Gomo

Reputation: 47

Why can HBase perform fast point queries while ClickHouse is not suitable

Based on the reason that "The key reason why point queries are expensive in ClickHouse is its sparse primary index of main MergeTree table engine family", ClickHouse is not good at point queries. However, HBase also uses LSM as the underlying implementation, but it is good at point queries. Why is this the case? I hope to get an explanation of this problem from the principles.

Upvotes: 0

Views: 274

Answers (1)

jsc0218
jsc0218

Reputation: 482

From the data structure's point of view, HBase uses pure lsm-tree, while ClickHouse organizes data parts into sth similar to lsm-tree, (there is no concept of levels in CH). But note that each part (or SST table in HBase) in CH are separated stored for each column, to optimize for batch reading.

Upvotes: 1

Related Questions