Let me know if i can optimize this query?

Question

does anybody have any suggestions to optimize this query :

SELECT COUNT(client_id) FROM (
  SELECT client_id FROM `cdr` WHERE 
    (DATE(start) BETWEEN '2014-04-21' AND '2014-04-25') AND 
    `service` = 'test' 
    GROUP BY client_id  
    HAVING SUM(duration) > 300 
)as t1

The problem is , inner query scans millions of rows and returns thousands of rows and it makes main query lazy.

Thanks.

Answer 1

Try below query, avoid functions in searching as it does not use index and kill the performance. "start" columns must be indexed.

SELECT COUNT(client_id) FROM (
  SELECT client_id FROM `cdr` WHERE 
    (start BETWEEN '2014-04-21 00:00:00' AND '2014-04-25 23:59:59') AND 
    `service` = 'test' 
    GROUP BY client_id  
    HAVING SUM(duration) > 300 
)as t1

Answer 2

Why not this?I have read somewhere that comparing dates directly with < or > works faster than Between.

SELECT Count(client_id) FROM `cdr` 
WHERE DATE(start) >=  '2014-04-21' 
    AND DATE(start) <= '2014-04-25'
    AND  `service` = 'test' 
GROUP BY client_id  
HAVING SUM(duration) > 300 

What was logic behind having sub query in your sql?

Answer 3

you can use direct query instead of sub query like as

SELECT COUNT(client_id) FROM `cdr` 
WHERE (DATE(start) BETWEEN '2014-04-21' AND '2014-04-25') 
    AND `service` = 'test' 
GROUP BY client_id  
HAVING SUM(duration) > 300