更新时间:2024-07-03 GMT+08:00
案例:改写SQL消除子查询(案例1)
现象描述
1 2 3 4 |
select 1, (select count(*) from customer_address_001 a4 where a4.ca_address_sk = a.ca_address_sk) as GZCS from customer_address_001 a; |
此SQL性能较差,查看发现执行计划中存在SubPlan,具体如下:
优化说明
此优化的核心就是消除子查询。分析业务场景发现a.ca_address_sk不为null,那么从SQL语义出发,可以等价改写SQL为:
1 2 3 4 5 |
select count(*) from customer_address_001 a4, customer_address_001 a where a4.ca_address_sk = a.ca_address_sk group by a.ca_address_sk; |
![](https://support.huaweicloud.com/distributed-devg-v2-gaussdb/public_sys-resources/note_3.0-zh-cn.png)
为了保证改写的等效性,在customer_address_001. ca_address_sk加了not null约束。
父主题: 实际调优案例