Since you're on a very very old version....your options are limited.

Things you could look at:

1) dbms_parallel_execute. Lets you have lots of concurrent deletes all running at the same time. This speeds up the job, but of course, works your server a lot harder.

2) manual implementation of what dbms_redefintion does, ie

- create an mview log on your source table to capture changes
- create table new_table as select from old_table where date_col > sysdate-365;
- now take small outage where you
a) rename new_table to old_table
b) use the mview log to apply any outstanding changes

(2) is more work , but gives you the opportunity to look at partitioning as well, so you wont ever get into this problem again.

But I'd seriously look at upgrading first...because once you are on 19c (or anything above 12.2) you can simply do:

alter table T move online including rows where date_col > sysdate - 365

or even doing that *and* converting to partitioning at the same time. Here's a video showing that in action



which will run reasonably efficiently, but the table stays online during the entire process.