Re: Seq scan on 10million record table.. why?

> 1) Make all types the same
> 2) If you are using some narrow type for big_table (say, int2) to save
> space, you can force narrowing conversion, e.g. "b.key1=ds.key1::int2". Note
> that if ds.key1 has any values that don't fit into int2, you will have
> problems. And of course, use your type used instead of int2.
>
> Best regards, Vitalii Tymchyshyn
>

This fixed my problem :)
Thanks Vitalii!

For the other suggestions made from Gabriele, unfortunately I can't
make an accurate data-partitioning 'cause (obviously) it will be quite
a big work and the customer finished the budget for this year, so
unless I choose to work for free... ;)

--
Vincenzo.