3 ms·
I remember reading that Clickhouse is quite bad at joins, which can be important if you have to build a snowflake schema. Is that still true? Is this something
by polskibus 5y ago
I remember reading that Clickhouse is quite bad at joins, which can be important if you have to build a snowflake schema. Is that still true? Is this something TimescaleDB would be better at?
- bvrmn 5y agoIf you data to join looks like not very huge dictionaries[1] (locations, types, etc) then ClickHouse can show amazing speeds. I had no any problems with a speed of usual joins though. [1] https://clickhouse.com/docs/en/sql-reference/dictionaries/ https://clickhouse.com/docs/en/sql-reference/dictionaries/
- PeterZaitsev 5y agoIn this case PostgreSQL may be able to come to the rescue :) There is Clickhouse FDW for PostgreSQL which in some cases can provide great speed with full join support https://github.com/adjust/clickhouse_fdw https://github.com/adjust/clickhouse_fdw
- Shicholas 5y agoyes, to take this a bit further, I love the idea that no matter what db is best, Postgres can be the starting point for all queries.