Our Goal is we need to minimize collect sats time:
we do use Combination for Full / Sample stats based on Row size:
- Smaple 50% for all table's statitics combination for tables > 5 Billion Rows
- FULL for all table's statitics combination for tables < 5 Billion Rows
While this still take long time and we came accross this idea, to minimize it
- we will not evaluate full or sample bu total row count, but for the number of values in each stats combination
Ex:
↧