Greg Smith wrote:
> Enterprise grade doesn't mean anything. Partitioning designs that
> require thousands of child tables to work right are fundamentally
> misdesigned anyway, so there is no reason for any of the contributors
> to the project to work on improving support for them. There are far
> too many obvious improvements that could be made to PostgreSQL, ones
> that will benefit vastly more people, to divert resources toward
> something you shouldn't be dong anyway like that.
>
my sql developer, who's been doing oracle for 15+ years, says postgres'
partitioning is flawed from his perspective because if you have a
prepared statement like..
SELECT fields FROM partitioned_table WHERE primarykey = $1;
it doesn't optimize this very well and ends up looking at all the
sub-table indicies. ir you instead execute the statement
SELECT fields FROM parritioned_table WHERE primarykey = constant;
he says the planner will go straight to the correct partition.
i haven't confirmed this for myself.