Skip to content

Comment on Challenges in join optimizationparent

Comments

Group join in academia generally points to having GROUP BY and one join in the same operation (since it's common to having aggregation and at least one join on the same attribute(s)). But just making a hash table on each side doesn't really do anything in itself (although making it on _one_ side is the typical start of a classic hash join); in particular, once you want to join on different keys, you have to regroup.

AboutSource Built by g1lg1l

Hackerly is an independent reader for Hacker News, built on the public HN API. Not affiliated with Y Combinator.