Not always. I use them to force hash joins. This way you can filter table A on an index, and filter table B on an index, then join them with a hash join. In

Not always. I use them to force hash joins. This way you can filter table A on an index, and filter table B on an index, then join them with a hash join. In a straight join postgres would try to use the same index for all of the joins. At least that was my experience. Maybe I'm not understanding the docs properly or using the correct terminology. I know that the end result was 10x faster than any index I could come up with.