案例:改写SQL消除in-clause
现象描述
in-clause/any-clause是常见的SQL语句约束条件,有时in或any后面的clause都是常量,类似于:
1
|
select count(1) from calc_empfyc_c1_result_tmp_t1 where ls_pid_cusr1 in (‘20120405’, ‘20130405’); |
或者
1
|
select count(1) from calc_empfyc_c1_result_tmp_t1 where ls_pid_cusr1 in any(‘20120405’, ‘20130405’); |
但是也有一些如下的特殊用法:
1
|
SELECT ls_pid_cusr1,COALESCE(max(round((current_date-bthdate)/365)),0)FROM calc_empfyc_c1_result_tmp_t1 t1,p10_md_tmp_t2 t2WHERE t1.ls_pid_cusr1 = any(values(id),(id15))GROUP BY ls_pid_cusr1; |
其中,id、id15为p10_md_tmp_t2中的两列,“t1.ls_pid_cusr1 = any(values(id),(id15))”等价于“t1.ls_pid_cusr1 = id or t1.ls_pid_cusr1 = id15”。
因此join-condition实质上是一个不等式,这种非等值的join操作必须使用nestloop连接,对应执行计划如下:
优化说明
测试发现由于两表结果集过大,导致nestloop耗时过长,超过一小时未返回结果,因此性能优化的关键是消除nestloop,让join使用更高效的hashjoin。从语义等价的角度消除any-clause,SQL改写如下:
1
|
selectls_pid_cusr1,COALESCE(max(round(ym/365)),0)from( ( SELECT ls_pid_cusr1,(current_date-bthdate) as ym FROM calc_empfyc_c1_result_tmp_t1 t1,p10_md_tmp_t2 t2 WHERE t1.ls_pid_cusr1 = t2.id and t1.ls_pid_cusr1 != t2.id15 ) union all ( SELECT ls_pid_cusr1,(current_date-bthdate) as ym FROM calc_empfyc_c1_result_tmp_t1 t1,p10_md_tmp_t2 t2 WHERE t1.ls_pid_cusr1 = id15 ))GROUP BY ls_pid_cusr1; |
优化后的SQL查询由两个等值join的子查询构成,而每个子查询都可以走更适合此场景的hashjoin。优化后的执行计划如下
优化后,从超过1个小时未返回结果优化到7s返回结果。