Re: Can I combine TEMP and RBS tablespace?

"Foggy" <sjefke12_at_hotmail.com> wrote in message news:f78e018a.0207160635.6ef96fde_at_posting.google.com...
> "Howard J. Rogers" <dba_at_hjrdba.com> wrote in message
news:<agvaj5$ljb$1_at_lust.ihug.co.nz>...
> > "Edzard" <edzard_at_volcanomail.com> wrote in message
> > news:5d75e934.0207151039.59ba72a8_at_posting.google.com...
>
> <SNIP>
>
> > Which is probably bad news. The extent sizes for temporary segments need
to
> > be a multiple of your sort_area_size plus one block. The extent sizes
for
> > your rollback segments probably ought to be something else - a common
> > multiple of 5 blocks and your multiblock_read_count.
> >
> <SNIP>
>
> Two small questions:
>
> - is it '(multiple of your sort_area_size) plus one block' or
> 'multiple of (your sort_area_size plus one block)' ?
>

Good question. It's the first one.... generate your multipole, and then add on a block to account for the segment header block. And since segment header blocks are only created in the initial extent of a segment, it's only INITIAL that's affected. NEXT should just be the multiple-of-sort_area_size.

And all of that is only true when you create temporary tablespaces using dictionary-managed datafiles (and I believe our original poster was using 7.3.4, so he would have been). If you're using locally-managed TEMPFILES, then the 'plus one block' is not needed, even for INITIAL.

Regards
HJR
> - Why is it 'plus one block'?
>
> Regards,
>
> Johan
Received on Tue Jul 16 2002 - 15:23:31 CDT