Re: Creating indexes takes too long

"Richard Foote" <richard.foote_at_bigpond.com> wrote in message news:<Jh_ba.2361$LT.7096_at_newsfeeds.bigpond.com>...
> "Hari Krishna Dara" <hari_newsgroups_at_yahoo.com> wrote in message
> news:16dcd029.0303121556.36a5ce88_at_posting.google.com...
> Hi Hara,
>
> The biggest overheads in creating an index are the read of the base table
> and the subsequent sorting involved.
>
> With regard to reading the table, the parameter
> db_file_multiblock_read_count is paramount (see archives for much discussion
> on this, also see Steve Adam's site www.ixora.com.au for good info). You may
> also want to look at parallel options and ensure the base table is suitably
> "striped" to minimize I/O related bottlenecks.
>
> With regard to the sort, ensure you have a healthy sized sort_area_size to
> minimize a disk sort if possible (which can be set at the session level).
> Or better still, look at the pga_aggregate_target if on 9i.
>
> Considering the size of your tables, I think there's definitely scope for
> improvement ...
>
> Good Luck
>
> Richard

Hello Richard,

Thanks a lot for all your suggestions, we will start looking at them.

Hari Received on Mon Mar 17 2003 - 16:49:29 CST