Question

/ * 大家可以帮助我如何优化下面提到的查询。在子查询中，我计算出货日期和（产品的）创建日期之间的差异为time_to_shift。然后在外部查询中我想基于time_to_shift slabs计算订单。下面给出的查询有效，但需要花费很多时间。任何人都可以告诉我如何在子查询上创建索引并在外部查询中使用它们。这也可以解决问题。 * /

SELECT DATE_FORMAT(b.ship_date,'%Y-%m')  AS MONTH, 
COUNT(*) AS Total_units_shipped, 
ROUND(100*COUNT(CASE WHEN time_to_ship <= 0 THEN time_to_ship END)/COUNT(*),2) AS '0 day',
ROUND(100*COUNT(CASE WHEN time_to_ship =1 THEN time_to_ship END)/COUNT(*),2) AS '1 day',
ROUND(100*COUNT(CASE WHEN time_to_ship> 1  AND time_to_ship<= 2 THEN time_to_ship END)/COUNT(*),2) AS '1-2 days',
ROUND(100*COUNT(CASE WHEN time_to_ship >2 AND time_to_ship <= 4 THEN time_to_ship END)/COUNT(*),2) AS '3-4 days',
ROUND(100*COUNT(CASE WHEN time_to_ship > 4 AND time_to_ship <= 10 THEN time_to_ship END)/COUNT(*),2) AS '5-10 days', 
ROUND(100*COUNT(CASE WHEN time_to_ship > 10 AND time_to_ship <= 20 THEN time_to_ship END)/COUNT(*),2) AS '11-20 days', 
ROUND(100*COUNT(CASE WHEN time_to_ship >20 THEN time_to_ship END)/COUNT(*),2) AS '>20 days' 
FROM(
   SELECT 
   DATE(fsp.sp_date_shipped) AS ship_date,
   CASE WHEN WEEKDAY(DATE(fso.soi_date_created)) = 5 THEN 
    (DATEDIFF(DATE(fsp.sph_date_shipped),DATE(fso.soi_date_created)) -1   ) 
      ELSE
       (DATEDIFF(DATE(fsp.sph_date_shipped),DATE(fso.soi_date_created)) ) 
    END
      AS Time_to_Ship 
    FROM dwh.f_suborders_oms fso 
    JOIN dwh.f_shipping_package fsp ON fso.soi_shipping_package_id=fsp.sp_shipping_package_id
    WHERE    fsp.sp_date_shipped BETWEEN DATE_FORMAT(DATE_SUB(CURDATE(),INTERVAL 1 MONTH ),'%Y%m01000000')
    AND DATE_FORMAT( CURDATE(),'%Y%m01000000')
 )b  
GROUP BY 1
LIMIT 1000000 ;

Answer 1

fsp.sp_date_shipped列类型是什么？看起来问题是

BETWEEN DATE_FORMAT(DATE_SUB(CURDATE(),INTERVAL 1 MONTH ),'%Y%m01000000')
AND DATE_FORMAT( CURDATE(),'%Y%m01000000').

尽量避免DATE_FORMAT并使用日期rathar而不是格式化输出让索引工作而不是完全扫描

优化具有子查询的MYSQL查询

1 个答案: