11 millions records : more than 30 seconds to retrieve simple request

[Funkiki]

Hi

I have a doubt about MySQL performance on large dataset

Hereafter my simple test table

`variable` varchar(80) NOT NULL,
`log_time` datetime NOT NULL DEFAULT '0000-00-00 00:00:00',
`milli` int(10) unsigned NOT NULL DEFAULT '0',
`num_value` double NOT NULL,
PRIMARY KEY (`variable`,`log_time`,`milli`) USING BTREE,
The engine is MyISAM. MySQL version is 5.5.8

I'm using the MySQL Query browser tool to make the performances tests.
The table contains 11 millions records.

When I run the following request :
SELECT * FROM my_table WHERE variable='a variable name' AND log_time >= '2010-10-11' AND log_time < '2010-10-25'
it takes more than 30 seconds to get the result that contains about 60000 records.
(of course, then, if I execute the request again, the answer is immediate , because of caching).

Now, if I do the following request :
SELECT variable,log_time,milli FROM my_table WHERE variable='another variable name' AND log_time >= '2010-10-11' AND log_time < '2010-10-25',
the result is less than 1second, but of course, the 'num_value' is missing.
If I run the same request with adding the 'num_value' field in the SELECT statement, then the running time is again about 30 seconds (and more)

Have you got an explanation about this ?

Why the first request takes so much time ?

Maybe the request with 'SELECT variable,log_time,milli' is very fast, because all the fields are part of the PRIMARY key and the process doesn't need to extract more data elsewhere ?
I'm currently rebuilding my table by adding the num_value field as member of the PRIMARY key to check that hypothesis. But, of course, this is not the good solution, because I don't want to make all the table fields part of the PRIMARY key, which would be stupid !
This is now more than 2 hours that the table is rebuilding... I will give you the result of this test when it will be finished.

But, I would like some clarification about that first test.

BR