 |
 |
Archives of the TeradataForum
Message Posted: Tue, 29 Apr 2003 @ 14:38:18 GMT
| Subj: | | Re: Collect statistics.. |
| |
| From: | | Fred W Pluebell |
I agree with some of the other comments - first make sure you really need updated statistics that often. Except for single-column indexes
on fields like date, timestamp, sequence number you seldom need new stats after every update run.
If you have plenty of (spool) space, have you tried updating existing statistics via COLLECT STATS without specifying COLUMN or INDEX?
Teradata can build multiple spool files in a single scan of the base table. If it's a new table, tell Teradata to COLLECT STATS on
individual COLUMNs (and PI if desired) while it's empty; then load the table, then re-collect. Or run multiple COLLECT STATS jobs in
parallel. In current releases, COLLECT STATS only takes a rowhash write lock for a very short time at the end.
| |
