Is this a one-off process? Or will you want to keep removing the oldest rows in the future?

If it's a one-off then I'd look at doing something like:

create table temp as 
  select <rows you want to keep>

<create indexes, constraints, etc. on temp>

alter table current rename to backup;

alter table temp rename to current;

This means you're only loading the data once instead of twice. And gives you a fast rollback if there are any problems - rename the tables back.

If this will be an ongoing process I'd partition the table too. Then remove the oldest partitions as necessary.

Either way, as a first step I'd look at upgrading to 19c. Not only is 11.2 now end-of-life, but this also opens up more (online) options which may help, such as:

- Filtered table move (alter table ... move including rows where ... )
- Single statement table partitioning (alter table ... partition ...)

For more on removing lots of rows fast, see https://blogs.oracle.com/sql/post/how-to-delete-millions-of-rows-fast-with-sql