RE: 10g INDEX_RS ? Syntax problem maybe?

From: <Christopher.Taylor2_at_parallon.net>
Date: Wed, 5 Dec 2012 10:02:35 -0600
Message-ID: <F05D8DF1FB25F44085DB74CB916678E88567D478C6_at_NADCWPMSGCMS10.hca.corpad.net>



I may have answered my own question - you may not be able to use an INDEX_RS with a HASH JOIN in which case the point is mute but I'd be interested in hearing other suggestions/thoughts/ideas (that are constructive). Regards,

Chris

-----Original Message-----

From: oracle-l-bounce_at_freelists.org [mailto:oracle-l-bounce_at_freelists.org] On Behalf Of Chris Taylor Sent: Wednesday, December 05, 2012 9:57 AM To: oracle-l_at_freelists.org
Subject: 10g INDEX_RS ? Syntax problem maybe?

From here, I'd like to get an INDEX_RS instead of a FAST FULL SCAN to only pull back the ID rows I want instead of examining all 24 M rows in the index.

I can accomplish this by modifying a SQL profile from q'[USE_HASH(_at_"SEL$5DA710D3" "MACD"@"SEL$1")]' to q'[USE_NL(@"SEL$5DA710D3" "MACD"@"SEL$1")]'.

Using the USE_NL instruction the INSERT time comes down from 2 minutes to 7 seconds.

I'm attempting to change that back to USE_HASH but get an INDEX_RS instead of an INDEX FAST FULL SCAN, but my hint is not working.

Here is my HINT:
INDEX_RS(MACD, MON_ACCOUNT_CHG_DET_IDX01)
--

http://www.freelists.org/webpage/oracle-l Received on Wed Dec 05 2012 - 17:02:35 CET

Original text of this message