If you have 12 rows inserted every second over 24hours of the day, when and where would you find a window to recreate the table.
Before you recreate the table you must analyze the impact on dependent objects (views, stored procedures / packages) etc.
AND you need to devise a method to generate the report you want -- the Row SCN might *not* suffice to meet your requirements.
Hemant K Chitale
You've already, in your other post on this subject, been given the simplest, least invasive solution to this.
dbms_scheduler job hourly, inserting cumulative values to a summary table.
Querying hourly increments from this is simple using analytics (hint ... look up lead / lag functions), or with a self join.
A slightly more complex query in the job could insert the increment instead.
And then, you also start to accumulate some historical data as well