Indianrock (2/11/2016)
Management's main concern is no blocking/locking.
That's where following the clustered index order comes to play.
You lock the continuous chunk of data and remove it.
Minimal number of pages affected, all done in a single pass.
When with un-ordered 1 record removal you update the same page as many times as many rows fit that page, and every time you delete a single row from a page every index needs to be updated, which means - several pages of it will locked with an exclusive lock.
And the same pages in all indexes will be locked as many times as many rows you delete from a single data page.