Oracle 19 skewed data and statistics gathering
Database version: Oracle 19 (latest CPU patch v 19.25)
2-node RAC - ASM storage
no license for partitioning
I have a large table (rowcount > 3 trillion rows or 3e9 rows).
The table has EVENTS related to CUSTOMERS
There is an index on event_date (DATE) and another index on customer_id (NUMBER).
We have appoximately 30 million different customers (3e7)
We have data over 20 years of events, i.e. starting around 2000.
The data skew is: some customers would have only very few events over the 20 year range. Some other "big" customers have millions of events over the 20 year range.
The number of "big customers" is in the hundreds, less than a 1000. But not just one or two. The ids for the "big" customer are spread throught the range of customer ids, they are not in a small consecutive group. But we can gather a list of all the "big" customers.