Yes it helped in that queries against A are now about 2 to 2.7 times slower than equivalents against B, while they used to be 3-4 times slower. It is not entirely satisfying though given that table A is almost twice smaller than table B, so I was hoping queries against A to be somewhat faster than against B (and this is the case when querying against both tables exported to a new DB). I guess I will have to move that table to a separate database file, so that it can be vacuumed regularly on its own. Thanks!