simonvandel commented on issue #8393: URL: https://github.com/apache/arrow-datafusion/issues/8393#issuecomment-1836270047
I just noticed that what I really want is to actually do a RIGHT join. That is, if there is no matching pricing for a timestamp, it should give null. Changing the query to that, Datafusion is much faster. I believe it's because with a RIGHT join, pricing becomes the outer table (single partition), while timestamps becomes the inner table (unspecified partitioning), which allows for greater parallelism (see https://github.com/apache/arrow-datafusion/blob/e19c669855baa8b78ff86755803944d2ddf65536/datafusion/physical-plan/src/joins/nested_loop_join.rs#L72-L77C4) But I think the issue should still be open - the LEFT join is still slower -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected]
