How can I speed up this MySQL query

Question

Can someone help me with this query? It's working fine but it take too much time. I determined the problem was too many OR statements in tb_activities . Does anyone have any ideas to make this query faster?

SELECT tb_posts.*
FROM   tb_posts
WHERE  EXISTS (SELECT c_id
               FROM   tb_users
               WHERE  tb_posts.c_uid = tb_users.c_id
                      AND tb_users.c_tokens > 0)
       AND NOT EXISTS (SELECT c_id
                       FROM   tb_activities
                       WHERE  tb_posts.c_url = tb_activities.c_url
                              AND tb_activities.c_category = 'gplus'
                              AND ( tb_activities.c_uid LIKE '%,6x1,%'
                                     OR tb_activities.c_uid LIKE '%,6x1'
                                     OR tb_activities.c_uid LIKE '6x1,%'
                                     OR tb_activities.c_uid = '6x1' ))
       AND NOT EXISTS (SELECT c_id
                       FROM   tb_blacklist
                       WHERE  tb_posts.c_url LIKE Concat('%', tb_blacklist.c_url
                                                  , '%')
                              AND tb_blacklist.c_times > 2
                              AND tb_blacklist.c_category = 'gplus')
       AND tb_posts.c_category = 'classic'
       AND tb_posts.c_status = 'run'
       AND tb_posts.c_nogplus = 0
GROUP  BY tb_posts.c_url
ORDER  BY tb_posts.c_cost DESC,
          tb_posts.c_date DESC
LIMIT  30  

Answer 1

What I did here is rewriting the first WHERE EXISTS ... . Why? Here is the answer. Another interesting read is this . So you might consider rewriting the query further. Unfortunately I don't have any more time now. But the main performance boost you will get anyway by adding (compound) indexes . Put indexes on columns on which JOIN s are based or columns which are often used in the WHERE clause.

SELECT tb_posts.*
FROM   tb_posts
INNER JOIN tb_users ON tb_posts.c_uid = tb_users.c_id
WHERE  tb_users.c_tokens > 0
       AND NOT EXISTS (SELECT c_id
                       FROM   tb_activities
                       WHERE  tb_posts.c_url = tb_activities.c_url
                              AND tb_activities.c_category = 'gplus'
                              AND ( tb_activities.c_uid LIKE '%,6x1,%'
                                     OR tb_activities.c_uid LIKE '%,6x1'
                                     OR tb_activities.c_uid LIKE '6x1,%'
                                     OR tb_activities.c_uid = '6x1' ))
       AND NOT EXISTS (SELECT c_id
                       FROM   tb_blacklist
                       WHERE  tb_posts.c_url LIKE Concat('%', tb_blacklist.c_url, '%')
                              AND tb_blacklist.c_times > 2
                              AND tb_blacklist.c_category = 'gplus')
       AND tb_posts.c_category = 'classic'
       AND tb_posts.c_status = 'run'
       AND tb_posts.c_nogplus = 0
GROUP  BY tb_posts.c_url
ORDER  BY tb_posts.c_cost DESC,
          tb_posts.c_date DESC
LIMIT  30

Also you might want to read about EXPLAIN , so you know which indexes are used (or not used).

And as a sidenote, Peter Kiss´ tip about changing the order of the WHERE clause is nonsense. The query optimizer handles this anyway.