Cardinality on pipeline function

From: Sidney Chen <huanshengchen_at_gmail.com>
Date: Wed, 30 Oct 2013 10:56:34 +0800
Message-ID: <CAM_ddu8=3qUi5QsE3ZnYkHSqdovE3ai7ZGZ7prU4=DZA5X0Hvw_at_mail.gmail.com>

Hi list,
Anyway to make optimizer aware of the actual rows from the pipelined function, something like dynamic sampling? For below sql, I need to add cardinality hint to specify the approx rows from the TABLE operation, the default cardinality on pipeline function seems 8168 and usually does not match the actual rows.

        SELECT /*+ cardinality(t1 1)*/COUNT(1)
          INTO V_COUNT
          FROM TABLE(GET_TABLE_FUNC(:v0)) t1,
               t2
         WHERE t1.id = t2.id
           AND t2.BATCH_ID = :v1;

--

Regards
Sidney Chen

--

http://www.freelists.org/webpage/oracle-l Received on Wed Oct 30 2013 - 03:56:34 CET