Imagine that you want to join two tables, and filter by a column that comes from the table in the right side of the join. On ClickHouse® the query a bit different than what you’d do in other databases like Postgres, and it will result in a big performance improvement.

Let’s say one of the tables is this events table, with 100M rows:

And the other table is this products one, with ~2M rows

If you will always filter the result after making the join as in the query above, you don’t need to make a join at all. ClickHouse® saves data column-by-column, so filtering by the values in a column is a very fast operation. If you rewrite the query as follows, it would be just as fast. And you wouldn’t have create a Join table for it:

Tinybird lets you create real-time API endpoints on in minutes instead of hours of days, powered by ClickHouse®. We’re still in private beta, but if you want to try out product, create an account here.