Why Have REPLACE INTO and INSERT ... on DUPLICATE KEY UPDATE?

REPLACE INTO will actually perform a delete and then an insert, while INSERT … ON DUPLCIATE KEY UPDATE will perform an update (as the name suggests). I would think the latter would be faster. I have not done any performance testing between the two, but it only seems logical the update would be faster than the delete/insert. Please correct me if I’m wrong.

Since the two statements both end up with the same result, I’m not sure yet what the benefit of REPLACE into it.

Yes, I realize both are MySQL additions, so please don’t bother telling me they’re not ANSI standard.

References:

REPLACE INTO