As of now UPDATE does in-place update that might change min/max values of an extent updated. This affects data clustering reducing effect of min/max pre-calculated aggregates.
To reduce the negative effect there must be a way to avoid in-place update for a subset of table columns.
Here is the list of sub-tasks to achieve the goal:

• deliver a function to enable/disable in-place update for a column of a table
• change syscat to store the flag for in-place update for a column
• change DML processing path so that it conditionally[1] does in-place or out-of-place update

1. in-place update in the syscat controls this