Many thousands of partitions

Hello,

I have question regarding one of caveats from docs:
http://www.postgresql.org/docs/8.3/static/ddl-partitioning.html

"Partitioning using these techniques will work well with up to perhaps a
hundred partitions; don't try to use many thousands of partitions."

What's the alternative? Nested partitioning could do the trick? I have
milions of rows(numbers, timestamps and text(<4kb), which are frequently
updated and there are also frequent inserts. Partitioning was my first
thought about solution of this problem. I want to avoid long lasting
locks, index rebuild problems and neverending vacuum.
Write performance may be low if at the same time I will have no problem
selecting single rows using primary key(bigint).Partitioning seems to be
the solution, but I'm sure I will end up with several thousands of
automatically generated partitions.

Thanks

--
Regards,
Grzegorz