Originally posted by slimdave

How about turning on monitoring for the table? This works at the partition level, and would allow you to run DBMS_STATS.GATHER_SCHEMA_STATS() with the gather_stale option to get good stats as soon as the data is flowing in.

This is an 8.1.7 database. I think table monitoring is a 9i feature?


You might also try running DBMS_STATS.SET_TABLE_STATS() and DBMS_STATS.SET_INDEX_STATS() after your weekend analyze. You can setup expected statistics at the partition level for both table and index that way. I'm not sure how this interfaces with table monitoring, but if you got these forecasted stats close enough on sunday night, then you could defer gathering stale stats until Monday night.
Excellent Idea!