RE: Partitioning best practices

Jaromir

> It will be nice to distinguish between DOP of creation (rebuild) of
> an object and DOP for access of the object.

This is a good one. Even if an ALTER TABLE/INDEX could be a viable way in many cases...

> It will be nice to see the SQL of the slaves as in good old 9'er
> days!

Nope. I like the 10g PX execution plans. There exactly show what's going on...

> > Most systems using PX are I/O bound.
>
> Well, few in-memory hash joins will do the job:)

And if they are partition wise joins it is even better... Unfortunately I see too many physical db designs that doesn't take advantage of that...

Regards,
Chris

--
http://www.freelists.org/webpage/oracle-l