Replace into.. Values/select and insert INTO ... Values/select on duplicate update similarities and differences

Replace into.. Values/select and insert INTO ... Values/select on duplicate update similarities and differences?

MySQL5.5 environment, T-table structure:
+-------+---------+------+-----+---------+----------------+
| Field | Type | Null | Key | Default | Extra |
+-------+---------+------+-----+---------+----------------+
| ID | Int (11) | NO | PRI | NULL | auto_increment |
| C1 | CHAR (5) |     YES | |                NULL | |
| C2 | CHAR (5) |     YES | |                NULL | |
+-------+---------+------+-----+---------+----------------+

Initialize table data:
Insert into T (C1,C2) VALUES (' 1 ', ' 1 ');
Insert into T (C1,C2) VALUES (' 2 ', ' 2 ');
Insert into T (C1,C2) VALUES (' 3 ', ' 3 ');

General statement:
ID Column No primary key/unique
1.2 and insert INTO ... values/select ... Insert Data normally
2. Full table plus exclusive lock (x)

ID column has primary key/unique
1.replace into T (ID,C1) VALUES (1, ' 11 '), only the column C1 value is updated, C2 is set to NULL, and the processing is equivalent to the delete
From t where id=1 add insert into T (ID,C1) VALUES (1, ' 11 '); 2 sql, which c2 the default value, so that the actual
During use, you need to give all column names in the table to avoid data loss.
2.insert into T (ID,C1) VALUES (3, ' a ') on duplicate key update c2= ' 333 ', when the ID has data duplication conflict,
SQL after update (c2= ' 333 '), otherwise insert into T (ID,C2) VALUES (4, ' ") on duplicate key update
C2= ' 444 ', executing insert INTO T (ID,C2) VALUES (4, ' 44 '), ignoring update c2= ' 444 ';
3. Add exclusive locks to updated rows (x)

From http://www.tokutek.com/2010/07/why-insert-on-duplicate-key-update-may-be-slow-by-incurring-disk-seeks/
It is reported that the proposed use in the daily, replace into ... values/select
Reason: INSERT INTO ... on pulicate key update ..., inserting data slowly, insert process, through Table Primary/unique key
Check whether the need to insert a record exists in the table, in this process involves disk data scanning, which is slow cause, no duplicate conflict data, direct insertion;
In addition, there are also related to the holding of locks, the corresponding impact on the system's concurrency;