Re: switching to 'direct path read' should help prevent thrashing buffer cache ?

<p.santos000_at_gmail.com> wrote in message news:1167841629.716228.178910_at_n51g2000cwc.googlegroups.com...
>

> Maybe I misinterpreted what you said, but I was under the impression
> that
> an update to a table with a ROWNUM <=500000 will lock only the 500000
> affected rows... likewise a parallel update with a ROWNUM <=500000 will
> also
> lock those 500K records .. but you seem to indicate that it's not the
> case.
>

    If you do a parallel update on a non-partitioned table     and it actually runs as a parallel update, then the table     is locked exclusively by the co-ordinator.

> I'll have to retest, but I believe I saw tx row lock contention and I
> would have
> thought that if the whole table was locked, I would have seen a
> different wait
> event.
>

    This suggests that you weren't running the update in parallel,     only the 'select' part, and the co-ordinator was updating the     table serially. Did you remember to:

        alter session enable parallel dml;

-- 
Regards

Jonathan Lewis
http://jonathanlewis.wordpress.com

Author: Cost Based Oracle: Fundamentals
http://www.jlcomp.demon.co.uk/cbo_book/ind_book.html

The Co-operative Oracle Users' FAQ
http://www.jlcomp.demon.co.uk/faq/ind_faq.html