In distributed systems, there's even a broadcast hash join, in which the hash table is not distributed across shards and assigned to workers but copied in full

In distributed systems, there's even a broadcast hash join, in which the hash table is not distributed across shards and assigned to workers but copied in full to every worker. That way, the other side of the join need not be sharded and shuffled to align join keys across workers. This strategy can save a lot of time and network bandwidth when one side of the join is small enough to fit into a worker's RAM and the other side is staggeringly massive (e.g., logs). It lets the massive side be processed in place, as it's streamed from storage.