Using parallelism - for mass deletes
Hi
We have a few databases running 12.2.0.1 and 19.8
One of the projects we are working on is to delete massive amount of records from a few tables (using a DATE criteria i.e. delete all the records with date greater than a given date) each having billions of rows. For some reasons, CTAS and other faster methods are not an options for this particular case so we are stuck with using DELETE. We wanted to ask if using a degree of parallelism would help us accomplish this task faster?
We are going to drop all the indices before the DELETE and then recreate those back. So basically we are looking into different things that could speed things up for those massive deletes. If using parallelism would help and if we have 48 cores on the server, what degree of parallelism should we use / would give us maximum benefits? What is the criteria / formula for using DOP ?