使用merge into 来更新目标表的个别字段

--更新职业 5s
merge into cr_gr x using
(
select b.custid,c.new_grbh,b.occupation /*职业*/,nvl(d.new_bm,'90') new_occupation /*新职业*/ 
from cm002@to_db2 b
left join scdy_grzhbh_dzb c on b.custid=c.old_grbh 
left join scdy_tmp_zgzy_dzb d on trim(b.occupation)=d.old_bm
where occupation<>' '
) new_tab
on (x.grbh=new_tab.new_grbh )
when matched then
update set x.zhiye=new_tab.new_occupation;

采用

--通过构造临时表test_source，把前后的对照关系找清楚，并且可观测到
merge into test_target
using test_source
on (test_source.id = test_target.id)--注意，如果是多条件，则 on后边一定要有个括号，把各条件都括进来
when matched then update set test_target.name = test_source.name
when not matched then insert values(test_source.name,test_source.id);--最后一行可以省略

1、UPDATE或INSERT子句是可选的 
2、UPDATE和INSERT子句可以加WHERE子句 
3、在ON条件中使用常量过滤谓词来insert所有的行到目标表中,不需要连接源表和目标表 
4、UPDATE子句后面可以跟DELETE子句来去除一些不需要的行 

参考：https://blog.csdn.net/weixin_39734304/article/details/79182416

为了对比效果

采用其他的方式更新

1
update test_target a set a.name=
 (
 with b as (select id,name from test_source )
 select a.name from a  where a.id=b.id
 
 )

2、创建中间的临时表

3、主表加索引

4、update语句

update cr_gr a set (a.zhiye,a.zhichen,a.jylx)=(
select nvl(d.new_bm,'90'),nvl(b.techpost,'0'), nvl(d.new_bm,'90') from cm002@to_db2 b 
left join scdy_grzhbh_dzb c on b.custid=c.old_grbh 
left join scdy_tmp_zgzy_dzb d on  b.occupation=d.old_bm
where c.new_grbh=a.grbh);

 

 

 

 

 

测试语句


drop table  t_stat;
create table t_stat(a1 int,b1 int,C1 varchar2(20));
insert into t_stat select level,mod(level,1000), null from dual connect by level<=100000;
commit;
select count(1) from T_STAT t;--10万条数据

select * from t_stat;
---构造B表
drop table  t_ref;
create table t_ref(b1 int,C2  varchar2(20));
insert into t_ref select mod(level,1000),null from dual connect by level<=1000;
update t_ref set c2='能更' where b1<=500; 
update t_ref set c2='' where b1>500; 

commit;

------------------开始更新
--方法1：1.3S
merge into t_stat  s
using t_ref  t
on (t.b1 = s.b1)
when matched then update set s.c1 = t.c2;
----验证
select * from t_stat;
--清空
update t_stat set c1 = null;
commit;
------------方法2 update  ----2.7s
update t_stat a set a.c1=
 (
       select b.c2 from t_ref b  where a.b1=b.b1
 );



select * from t_ref