mysql 查询中 select 的 select 的优化

Question

我的 mysql 查询需要将近 7 秒来获取数据，其中我使用了 select 的 select。

SELECT s.id
     , s.user_id
     , COUNT(s.od_status) od_status
     , (SELECT count(t2.od_status) 
          FROM subscription t2 
         WHERE od_status <> 3 
           AND t2.od_status <> 7 
           AND t2.od_status <> 8 
           AND t2.user_id = s.user_id 
           AND DATE(IF(t2.rescheduling_delivery_date IS NULL, t2.dated, t2.rescheduling_delivery_date)) BETWEEN DATE_FORMAT('2021-07-26', '%Y-%m-01') AND '2021-07-26'
        ) od_status_count 
  FROM subscription 
  LEFT 
  JOIN users u
    ON u.id = s.user_id 
 WHERE DATE(IF(s.rescheduling_delivery_date IS NULL, s.dated, s.rescheduling_delivery_date)) BETWEEN DATE_FORMAT('2021-07-26' , '%Y-%m-01') AND '2021-07-26' 
 GROUP 
    BY s.user_id;

我需要优化上面的查询，我得到 7 秒的是在上面的查询 select of select 下方添加

(select count(t2.od_status) from subscription t2 where od_status<>3 and t2.od_status<>7 and t2.od_status<>8 
and t2.user_id=subscription.user_id and 
date(IF(t2.rescheduling_delivery_date IS NULL,t2.dated,t2.rescheduling_delivery_date)) 
between DATE_FORMAT('2021-07-26' ,'%Y-%m-01') AND '2021-07-26') as od_status_count

任何优化建议或其他使用方法。

注意：我已将 index 给 subscription user_id table

Answer 1

为了使优化器能够使用索引，您需要：

改写您的查询。使用 OR 或 UNION/UNION ALL.
删除 date() 函数并用时间戳比较代替它。

如果不这样做，您仍然可以通过为子查询添加索引来获得更好的性能。这是最简单的选择，但不会真正得到很大的改进。您可以添加以下索引：

create index ix1 on subscription (user_id, od_status);

为了进一步提高性能，您可以改用覆盖索引：

create index ix1 on subscription (
   user_id,
   od_status,
   rescheduling_delivery_date, 
   dated, 
   rescheduling_delivery_date
);

Answer 2

按照 @RahulBiswas 的建议，我将子查询更改为 LEFT JOIN 并且查询在 0.16 秒

内执行

select 
`subscription`.`id`, 
`subscription`.`user_id`, 
COUNT(subscription.od_status) as od_status ,
x.od_status_count
from `subscription` left join `users` on `users`.`id` = `subscription`.`user_id`
left join(
  select t2.user_id,count(t2.od_status)as od_status_count from subscription t2 where od_status<>3 and t2.od_status<>7 and t2.od_status<>8 
and 
date(IF(t2.rescheduling_delivery_date IS NULL,t2.dated,t2.rescheduling_delivery_date)) 
between DATE_FORMAT('2021-07-26' ,'%Y-%m-01') AND '2021-07-26'
group by t2.user_id)x
on x.user_id=subscription.user_id    
where date(IF(subscription.rescheduling_delivery_date IS NULL,subscription.dated,subscription.rescheduling_delivery_date)) 
between DATE_FORMAT('2021-07-26' ,'%Y-%m-01') AND '2021-07-26' 
group by `subscription`.`user_id`

我将子查询更改为如下连接，

left join(
select t2.user_id,count(t2.od_status)as od_status_count from subscription t2 where od_status<>3 and t2.od_status<>7 and t2.od_status<>8 
and 
date(IF(t2.rescheduling_delivery_date IS NULL,t2.dated,t2.rescheduling_delivery_date)) 
between DATE_FORMAT('2021-07-26' ,'%Y-%m-01') AND '2021-07-26'
group by t2.user_id)x
on x.user_id=subscription.user_id

mysql 查询中 select 的 select 的优化

optimization for select of select in mysql query

mysql

select

query-optimization