So Created_tmp_tables is being bumped by one, but Created_tmp_disk_tables is not.
Had I been leaving off the SQL_NO_CACHE, the query cache would have been catching repeated executions of the same query in testing and the counters would not have been moving at all except for the very first test.
By commenting, you are accepting the DISQUS terms of service.
The _server_ is the component truncating/setting the BLOB to null.
id=1 Simon Mudd is right, but there are several things that must come together to make this bad:1.The actual size of the column or its content do not matter - even a TINYTEXT that is empty is enough.Since the MEMORY storage engine cannot represent any TEXT or BLOB types at all, this forces My SQL to realize the table as an on-disk My ISAM table. Run show session status:[email protected] [test_world]2 rows in set (0.00 sec)As you can see the Created_tmp_tables counter increased by one (in My SQL 5.0 is increases by two because the SHOW STATUS itself creates an in-memory tmp table which is being counted).(Moving a lob column might also rebuild the lob index - I'm not sure). When you sign in to comment, IBM will provide your email, first name and last name to DISQUS.