带有子查询的慢速MySQL查询

Question

I'm experiencing a problem with this query in PHP. It is so slow I cannot connect to the server for few minutes after executing it. I've been removing separate subqueries away from main query and found out, that it works perfectly smooth, after I remove the last inner query.

Table "u" has 30.000 rows, table "u_r" around 17.000, table "u_s" around 13.000 and table "s" around 100. Even though tables "u_r" and "u_s" have lots of rows, only 2-3 rows have same "id_u" that matches the condition.

I hope, I provided enough information. If you need to know anything else, feel free to ask in the comments.

SELECT DISTINCT id_p
FROM u
WHERE
    '$x' IN(
        SELECT id_p
        FROM u_p
        WHERE id_u=u.id_u
    )
    AND
    (
        '$y' IN (
            SELECT id_r
            FROM u_r
            WHERE id_u=u.id_u
        )
        OR
        '$y' IN (
            SELECT DISTINCT id_r
            FROM s
            WHERE id_s IN (
                SELECT id_s //without this query, everything works fine
                FROM u_s
                WHERE id_u=u.id_u
            )
        )
    )

Answer 1

I think you can change:

SELECT DISTINCT id_r 
        FROM s 
        WHERE id_s IN ( 
            SELECT id_s //without this query, everything works fine 
                FROM u_s 
                WHERE id_u=u.id_u 

to

SELECT s.id_s
        FROM u_s 
             INNER JOIN s ON u_s=id_u=u.id_u
        WHERE u_s.id_u = u.id_u
        GROUP BY s.id_r

Not tested as I'm trying to wrap my head around the structure.

Also remember to check the indexes. As a rough guide, you should have anything in the "WHERE" and anything in the "ON" indexed:

u_p.id_u
u_r.id_u
u_s.id_u
s.id_s

Answer 2

Test that this produces the same results - should be much faster.

1. Replace the IN statements with inner joins. More opportunities for query optimiser to optimise this way
2. Eliminate your OR statement - I did this using a UNION. I think you could probably eliminate the OR using a LEFT JOIN instead if you wanted.
3. You'll need to test this - I don't have any sample data.

 select distinct id_p from u inner join u_p on u.id_u = u_p.id_u and u_p.id_u = '$x' inner join ( select id_r.id_u from u_r where id_r = '$y' union select id_r from s inner join u_s on s.id_u = u_s.id_u where id_r = '$y' ) as subq on u.id_u = subq.id_u