我正在尝试更新 Oracle 中的多个列,并在经历了一些场景后解决了下面的代码。
我的问题是,更精简的代码不会更新我需要的内容,而更复杂的代码会更新——我正在尝试通过最小的处理能力来实现这一点,因为我正在查看数十万甚至更多的更新。
所以第一个代码:
UPDATE table@database1 i2
SET (i2.access, i2.permission) =
(select
i2m.access, i2m.permission
from temporarytable ss
JOIN table2 pgm
ON ss.secgroup = pgm.string
JOIN table i2m
ON pgm.hmy = i2m.hgroup
AND ss.pername = i2m.sobjname
JOIN table2@database1 pg
ON ss.secgroup = pg.string
WHERE
pg.string = 'string' -- this limits the updates to a specific subset of data
and i2m.hmy > 0 -- this for some freak records in both tables that are missing a primary key
and pg.hmy = i2.hgroup -- this matches the key of the 'string' from above to the records i need to update
and ss.pername = i2.sobjname -- further condition on which records to update. so only the ones that match from the temp table to the target table
and ss.orig_hmy = i2.hmy) -- further condition to make sure i am updating only records matching between temp table and target table
现在,如果我运行它,而不是只更新大约 700 条与上述子查询匹配的记录,它会更新表“table@database1”中的所有记录,我不明白为什么(可能是我不明白的事情之一甲骨文 :) )
但是,如果我运行以下命令——唯一的区别是我将整个子查询插入到“存在的地方”中——那么这只会更新我需要的内容。我的问题是,按照我的理解,子查询运行了两次——一次在更新中,一次在 where 子句中——我会说这是对处理能力的浪费。
UPDATE table@database1 i2
SET (i2.access, i2.permission) =
(select
i2m.access, i2m.permission
from temporarytable ss
JOIN table2 pgm
ON ss.secgroup = pgm.string
JOIN table i2m
ON pgm.hmy = i2m.hgroup
AND ss.pername = i2m.sobjname
JOIN table2@database1 pg
ON ss.secgroup = pg.string
WHERE
pg.string = 'string'
and i2m.hmy > 0
and pg.hmy = i2.hgroup
and ss.pername = i2.sobjname
and ss.orig_hmy = i2.hmy)
where exists (select
i2m.access, i2m.permission
from temporarytable ss
JOIN table2 pgm
ON ss.secgroup = pgm.string
JOIN table i2m
ON pgm.hmy = i2m.hgroup
AND ss.pername = i2m.sobjname
JOIN table2@database1 pg
ON ss.secgroup = pg.string
WHERE
pg.string = 'string'
and i2m.hmy > 0
and pg.hmy = i2.hgroup
and ss.pername = i2.sobjname
and ss.orig_hmy = i2.hmy)
注意:如果它没有显示,我有多个具有相同架构的数据库。我正在尝试使用来自主模式的信息更新跨数据库的表。Temp 表充当不同且需要更新的记录的存储库 - 如果仅 15% 与主模式不同,则没有理由更新数百万条记录。