Waldomiro wrote:
> ...
> I´m thinking It hapens because in the 7 millions tables, the same 8k
> block has diferent records with different keys, so only a few records
> with 'my_key' is retrieved when I read a 8k block.
> In the part_table, all records stored in a 8k block have 'my_key', so
> It´s much optimized.
>
> My doubt, there is a way to defrag my 7 millions table to put all
> records with the same key in the same 8k block?
Read about the "CLUSTER ON index-name" SQL command. It does exactly what you're asking.
Craig