> I thought a sequential scan was slow because it was limited by disk I/O Not necessarily. Suppose you have a fairly big table that fits in the shared buffer c

> I thought a sequential scan was slow because it was limited by disk I/O Not necessarily. Suppose you have a fairly big table that fits in the shared buffer cache and you run a query with a predicate that returns only a handful of rows. Then you would be better off scanning parts of it in parallel and then combining the results.