Re: 9i multi cache buffer

There is no keep cache (nor a recycle cache) for any non-default block sizes (which is one reason, amongst many, why the use of non-standard block sizes is a very, very bad idea. They were invented to make tablespace transportation a practical possibility, and for no other reason. They are likely to play merry havoc with performance tuning, not improve things).

Therefore, your table A will always be housed in the 2K buffer (as sized by your db_2K_cache_size parameter).

There is a right block size for a database, and a wrong size. And it all depends on what operating system you are using. Therefore my strong advice is: unless you are transporting tablespaces, avoid this "feature" like the plague.

I quote this little snippet from Steve Adams' website (www.ixora.com.au):

Under Oracle9i individual tablespaces can have a block size that differs from the database block size. Of course, this should not be considered if the database is using buffered I/O, because in that case the database block size must match the file system buffer size exactly.

'Nuff said.

Regards
HJR "charlie cs" <cs3526_at_ureach.com> wrote in message news:qeZz9.5609$6Z.5205_at_nwrddc01.gnilink.net...
> I am kind of confused with 9i multi cache buffer concept.
> Say in my database,
> I have set up:
> Tablespace A: 2k block size,
> B: 8k block size (default)
> and I set
> db_2K_cache_size = 20m
> db_cache_keep = 20m
>
> If I create a table in tablespace A,
> then I alter table storage (buffer_pool keep), where will this table stay
in
> memory?
> db_2k_cache_size or
> db_cache_keep?
>
> Thanks for your help
>
>
>
Received on Mon Nov 11 2002 - 20:29:02 CST