Simple MySQL queries taking long time to compute

Question

I am just learning MySQL and I have a problem.

Sometimes for various queries MySQL starts calculating at 100% CPU usage for 15-20 seconds and than it returns the result normally saying:

Query took 0.1780 sec.

It happens on very simple queries. For example this query took 0.36 seconds.

(SELECT DISTINCT a1.actor 
 FROM   actors AS a1, 
        actors AS a2 
 WHERE  a1.title = a2.title 
        AND a1.YEAR = a2.YEAR 
        AND a1.actor = a2.actor 
        AND a1.character_name <> a2.character_name) 

The listing of the table (7000 rows) took 0.001 seconds.

On the other hand when I just want to combine these two, MySQL goes crazy and starts calculating for 30 seconds and then finally returning: Query took 0.1800 sec)

SELECT actor 
FROM   actors 
WHERE  actor NOT IN (SELECT DISTINCT a1.actor 
                     FROM   actors AS a1, 
                            actors AS a2 
                     WHERE  a1.title = a2.title 
                            AND a1.YEAR = a2.YEAR 
                            AND a1.actor = a2.actor 
                            AND a1.character_name <> a2.character_name) 

Why is this happening?

Here is an other example. This query takes around 2 second and reports 0.5

SELECT DISTINCT a1.character_name 
FROM   (actors AS a1 
        NATURAL JOIN movies AS m1), 
       (actors AS a2 
        NATURAL JOIN movies AS m2) 
WHERE  a1.character_name = a2.character_name 
       AND ( m1.title <> m2.title 
              OR ( m1.title = m2.title 
                   AND m1.year <> m2.year ) ) 
       AND m1.country <> m2.country 

On the other hand this query takes 15-20 seconds, CPU 100% but reports 0.3 seconds. (The only difference is a bracket after AND ( .... )

SELECT DISTINCT a1.character_name 
FROM   (actors AS a1 
        NATURAL JOIN movies AS m1), 
       (actors AS a2 
        NATURAL JOIN movies AS m2) 
WHERE  a1.character_name = a2.character_name 
       AND m1.title <> m2.title 
        OR ( m1.title = m2.title 
             AND m1.YEAR <> m2.YEAR ) 
           AND m1.country <> m2.country 

I am using phpMyAdmin and the latest XAMPP for testing.

Update:

The wrong query times seem to be related to phpMyAdmin, on command line I get the following times:

• 1st query: MySQL: 0.36 s - PostgreSQL: 0.37 s
• 2nd query: MySQL: 43 s - PostgreSQL: 0.42 s
• 3rd query: MySQL: 4.86 s - PostgreSQL: 0.05 s
• 4th query: MySQL: 1 min 5 s - PostgreSQL: 15 seconds

So I have the answer for why were the query times reported wrongly (bug either in phpMyAdmin or XAMPP), I am interested in why do such similar queries have such a big difference in running time?

Update 2:

Just for completeness I did the testing with PostgreSQL too

Answer 1

Have you tried testing your queries with mysql cmd prompt ??? If the problem still persists then the issue might have been with mysql but if the problem is solved then i think you have a problem with phpmyadmin. So let me know that whether your problem still persists after trying your queries with mysql cmd prompt.

Answer 2

Change this (Your first query)

SELECT DISTINCT a1.actor 
FROM   actors AS a1, 
actors AS a2 
WHERE  a1.title = a2.title 
AND a1.YEAR = a2.YEAR 
AND a1.actor = a2.actor 
AND a1.character_name <> a2.character_name) 

to this:

SELECT *
FROM actors a1
JOIN actors a2 ON (a1.title = a2.title AND a1.actor = a2.actor)
GROUP BY a1.actor
HAVING a1.character_name <> a2.character_name

and use the same style for the others, also make sure that you have proper indexes on your tables.

Answer 3

SELECT actor 
FROM   actors 
WHERE  actor NOT IN (SELECT DISTINCT a1.actor 
                     FROM   actors AS a1, 
                            actors AS a2 
                     WHERE  a1.title = a2.title 
                            AND a1.YEAR = a2.YEAR 
                            AND a1.actor = a2.actor 
                            AND a1.character_name <> a2.character_name)

The above query looks like it's trying to select actors who have never played multiple characters on a single title. You could have just said:

select   actor
from     actors
group by actor, year, title
having   count(character_name) = 1

However, I know that your question wasn't in regard to your sql writing abilities and you are just trying to figure out why the strange behaviour from MySql. My guess is that it is excluding certain things from the execution time. For example, when google says it took .09 seconds to get your results but you know you waited 10 seconds for the page to load. Google didn't account for the 9.91 seconds it was gonna take to get from its server to your computer... just how long it took for them to query for the data.

This definitely looks like a glaring problem that MySql should address though because unlike Google who can't know the other part of the equation, MySql should be able to incorporate the whole process into the time calculation.

Answer 4

Try to use EXPLAIN to profile your queries. And my advise - not to use subqueries.