DATETIME vs. INT(11)?

[yawnmoth]

I'm curious... what's the advantage of using DATETIME over INT(11)?

I ran some tests and near as I can tell, there isn't one. Indeed, it seems to me that there's actually a bit of a disadvantage.

The first test uses the following schema:

CREATE TABLE test (
   test DATETIME
);
INSERT INTO test VALUES ('2007-12-31 23:59:59'),('2010-11-30 23:59:59'),('2003-12-31 01:02:03');

I then run the following query:

EXPLAIN SELECT YEAR(test) AS `year`, MONTH(test) AS `month`
FROM test
GROUP BY `year`, `month`

Here's what I get:

id = 1
select_type = SIMPLE
table = test
type = ALL
rows = 3
Extra = Using temporary; Using filesort

Curious to see how it'll perform if test is a key, I do the following:

ALTER TABLE test ADD KEY(test);

Here's what I get:

id = 1
select_type = SIMPLE
table = test
type = index
key = test
key_len = 9
rows = 3
Extra = Using index; Using tempory; Using filesort

Now I try it with INT(11):

CREATE TABLE test2 (
   test INT(11)
);
INSERT INTO test2 VALUES (UNIX_TIMESTAMP('2007-12-31 23:59:59')),(UNIX_TIMESTAMP('2010-11-30 23:59:59')),(UNIX_TIMESTAMP('2003-12-31 01:02:03'));

I then run the following query:

EXPLAIN SELECT from_unixtime(test, '%Y') AS `year`, from_unixtime(test, '%m') AS `month`
FROM test2
GROUP BY `year`, `month`

Here's what I get:

id = 1
select_type = SIMPLE
table = test2
type = ALL
rows = 3
Extra = Using temporary; Using filesort

So, there's no difference between this and the last one. Same thing happens if I make test a key. Only difference then is that the key_len is 5 and not 9.

So I'm left to wonder... what, really, is the point to DATETIME? It doesn't seem to offer any advantages over INT(11) and indeed, using it yields a larger key_len, which is a clear disadvantage of it.