好得很程序员自学网

<tfoot draggable='sEl'></tfoot>

sql优化实战 把full join改为left join +union all(从5分钟降为10秒

今天收到一个需求,要改写一个报表的逻辑,当改完之后,再次运行,发现运行超时。

因为特殊原因,无法访问客户的服务器,没办法查看sql的执行计划、没办法知道表中的索引情况,所以,尝试从语句的改写上来优化。

一、原始语句如下:

?

1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

select isnull (vv.customer_id,v.customer_id) as customer_id,

         isnull (vv.business_date, replace (v.business_date, '-' , '' )) as business_date,

         v.prod_id,

         v.sales,

         vv.visit_count,

     v.all_sales

from

(

   SELECT a.customer_id ,

         max ( month )+ '-01' as business_date,

       a.PROD_ID ,

       SUM ( CAST (VALUE AS NUMERIC (38, 3))) sales,

       sum ( SUM ( CAST (VALUE AS NUMERIC (38, 3)))) over(partition by a.customer_id) as all_sales

                            

   FROM   TB_IMPORT_SALES a

   WHERE   a.customer_id IS NOT NULL

       AND a.PROD_ID IS NOT NULL

             and a. month = '2016-11'

   GROUP BY a.customer_id ,

       a.PROD_ID

)v

full join

(

   SELECT customer_id,

         max (a.business_date) as business_date,

       COUNT (*) AS VISIT_COUNT

     FROM TB_CALL_STORE a WITH (NOLOCK)

     inner join TB_TIME d

     on a.business_date = d.t_date

     where d. section = '2016-11'

     GROUP BY customer_id

)vv

on v.customer_id = vv.customer_id

原来是left join,虽然查询比较慢,但是2分钟能查出来,现在按照业务要求,需要看到所有数据,所以改成了full join,改了之后5分钟都查不出结果。

二、改写后的代码

?

1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

select v.customer_id,

         replace ( max (v.business_date), '-' , '' ) as business_date,

         v.prod_id,

         max (v.sales_volume) sales_volume ,

         max (v.visit_count) visit_count,

         max (v.all_sales_volume) all_sales_volume

from

(

   SELECT a.customer_id ,

         max (biz_month)+ '-01' as business_date,

       a.PROD_ID ,

       SUM ( CAST (VALUE1 AS NUMERIC (38, 8))) sales_volume,

       sum ( SUM ( CAST (VALUE1 AS NUMERIC (38, 8)))) over(partition by a.customer_id) as all_sales_volume,

             null as visit_count

                            

   FROM   TB_IMPORT_SALES a

   WHERE   a.customer_id IS NOT NULL

       AND a.PROD_ID IS NOT NULL

             and a. month = '2016-11'

   GROUP BY a.customer_id ,

        a.PROD_ID

   union all

 

   SELECT customer_id,

         max (a.business_date) as business_date,

           p.prod_id,

           null ,

           null ,

       COUNT (*) AS VISIT_COUNT

     FROM TB_CALL_STORE a WITH (NOLOCK)

     cross apply

     (

         select top 1 prod_id from TB_PRODUCT with (nolock)

     )p

     inner join TB_TIME d

     on a.business_date = d.t_date

     where d. section = '2016-11'

     GROUP BY customer_id,p.prod_id

)v

group by v.customer_id,

      v.prod_id

由于代码本身比较简单,没办法再进一步简化,而由于连接不了服务器,其他的方法也用不上,甚至没办法分析到底是什么导致运行这么慢。

想了想,full join 本质上就是 2次left join+union ,无非就是合并数据,于是尝试一下用union all来直接合并数据,现在改成unoin all最后,就不需要full join。

但是考虑到第2段代码中并没有prod_id这个字段,所以这里在第2段代码加上了cross apply随便取出一个产品的id,这样就有prod_id这个字段,可以合并了。

修改之后,果然速度降到了10多秒。

到此这篇关于sql优化实战 把full join改为left join +union all(从5分钟降为10秒)的文章就介绍到这了,更多相关left join +union all内容请搜索服务器之家以前的文章或继续浏览下面的相关文章希望大家以后多多支持服务器之家!

原文链接:https://blog.csdn.net/sqlserverdiscovery/article/details/53542543

查看更多关于sql优化实战 把full join改为left join +union all(从5分钟降为10秒的详细内容...

  阅读:17次