Archives of the TeradataForum
Message Posted: Mon, 22 Feb 2010 @ 14:59:35 GMT
| Subj: | | Re: Avoiding an all-row scan |
| |
| From: | | Dieter Noeth |
mjrodriguezp wrote:
| | - The table BDD_TABLAS.FIN_BORRAR_1 has only one row with the date I want to use SELECT * FROM BDD_TABLAS.FIN_BORRAR_1; | |
Statistics on FIN_BORRAR_1.fec_vigencia, so the optimizer knows about that single row? If not it will estimate the number of rows equal to the
number of AMPs in your system.
Statistics on PCL_HIS_CONTRATOS: fec_vigencia/PARTITION?
Is that all-row-scan really a full-table-scan or "enhanced by dynamic partition elimination"?
| | The thing is than I need to have the date in a field of a table like this because it can change. | |
| | Is there a way to optimize the second or third query using the table FIN_BORRAR_1? | |
If that table is always a single row you might try a view:
REPLACE VIEW FIN_BORRAR_1 AS
SELECT DATE '2009-12-31' AS fec_vigencia;
In TD12 the optimizer will completely remove the view-access and create exactly the same plan as the first query.
Btw, don't write dates using the old integer format anymore, better use "DATE '2010-02-22'"
Dieter
| 
