ParadeDB speeds up multi-table Postgres search joins by up to 1000x
ParadeDB says it has built native support for 'search joins'—relevance-ranked search queries across multiple normalized Postgres tables—without requiring data to be denormalized into a separate system like Elasticsearch. Using a StackOverflow dataset benchmark spanning users, posts, comments and badges tables, the company reports a three-table Top-K search query dropped from about 2 minutes on vanilla Postgres to roughly 100 milliseconds.