Question

I have the following SQL query:

select expr1, operator, expr2, count(*) as c  from log_keyword_fulltext  group by expr1, operator, expr2  order by c desc limit 2000; 

Problem: The count(*) as part of my order by is killing my application, probably because it don’t use index. I would like to know if there is any way to make it faster, like for example a select inside of another select, or something like that.

My SELECT explained:

+----+-------------+----------------------+-------+---------------+-------+---------+------+--------+----------------------------------------------+ | id | select_type | table                | type  | possible_keys | key   | key_len | ref  | rows   | Extra                                        | +----+-------------+----------------------+-------+---------------+-------+---------+------+--------+----------------------------------------------+ |  1 | SIMPLE      | log_keyword_fulltext | index | NULL          | expr1 | 208     | NULL | 110000 | Using index; Using temporary; Using filesort |  +----+-------------+----------------------+-------+---------------+-------+---------+------+--------+----------------------------------------------+ 

UPDATED:

I tried to do a subquery like that

select * from (select b.expr1,b.operator,b.expr2,count(*) as c  from log_keyword_fulltext b group by b.expr1,b.operator,b.expr2) x  order by x.c desc limit 2000; 

its working but not faster, following is the explain:

+----+-------------+------------+-------+---------------+-------+---------+------+--------+----------------+ | id | select_type | table      | type  | possible_keys | key   | key_len | ref  | rows   | Extra          | +----+-------------+------------+-------+---------------+-------+---------+------+--------+----------------+ |  1 | PRIMARY     | <derived2> | ALL   | NULL          | NULL  | NULL    | NULL |  38398 | Using filesort |  |  2 | DERIVED     | b          | index | NULL          | expr1 | 208     | NULL | 110000 | Using index    |  +----+-------------+------------+-------+---------------+-------+---------+------+--------+----------------+ 

You can check that now, its not using temporary anymore, but it still with the same performance. any recommendation ?

Answer 1

What am I missing? I don’t see a WHERE clause. It looks to me you’re requesting a table scan.

If you are counting on your ‘LIMIT’ clause, you’re out of luck – that’s the COUNT aggregate calculation.