提高存储过程性能

Question

我是一个非数据库家伙..你可以请你分享你的意见，以改善下面的sp样本，如使用全局临时表或索引或可能会改善现有的查询。在原代码中，我会对不同的桌子有很多更新查询。谢谢！

CREATE OR REPLACE PROCEDURE SYSTEM.process_log1 
IS 
    cursor cur_attuid_change is 
select 
    sup.id as id, 
    sup.name as name, 
    sup.role as role, 
    sup.technology as technology 
from main hr, secondary sup 
where sup.id=hr.id; 

BEGIN 
    -- update records in main tables from the cursor 
    for rec_attuid_change in cur_attuid_change loop 

    update main t1 
    set t1.id = rec_attuid_change.id, 
      t1.name = rec_attuid_change.name, 
      t1.ROLE=rec_attuid_change.role, 
      t1.technology=rec_attuid_change.technology 
    where t1.id = rec_attuid_change.id; 
    commit; 

    update main1 t1 
    set t1.id = rec_attuid_change.id, 
      t1.name = rec_attuid_change.name, 
      t1.ROLE=rec_attuid_change.role, 
      t1.technology=rec_attuid_change.technology 
    where t1.id = rec_attuid_change.id; 
    commit; 

    update main2 t1 
    set t1.id = rec_attuid_change.id, 
      t1.name = rec_attuid_change.name, 
      t1.ROLE=rec_attuid_change.role, 
      t1.technology=rec_attuid_change.technology 
    where t1.id = rec_attuid_change.id; 
    commit; 
    end loop; 


END; 

/

Answer 1

试试这个：

CREATE OR REPLACE PROCEDURE SYSTEM.process_log1 

BEGIN 

    update main t1 
    set (t1.id, t1.name, t1.ROLE, t1.technology) = 
     (select 
     sup.id as id, 
     sup.name as name, 
     sup.role as role, 
     sup.technology as technology 
     from 
     main hr, 
     secondary sup 
     where 
      sup.id=hr.id 
      and sup.Id = t1.Id); 
    commit; 

    update main1 t1 
    set (t1.id, t1.name, t1.ROLE, t1.technology) = 
     (select 
     sup.id as id, 
     sup.name as name, 
     sup.role as role, 
     sup.technology as technology 
     from 
     main hr, 
     secondary sup 
     where 
      sup.id=hr.id 
      and sup.Id = t1.Id); 
    commit; 

    update main2 t1 
    set (t1.id, t1.name, t1.ROLE, t1.technology) = 
     (select 
     sup.id as id, 
     sup.name as name, 
     sup.role as role, 
     sup.technology as technology 
     from 
     main hr, 
     secondary sup 
     where 
      sup.id=hr.id 
      and sup.Id = t1.Id); 
    commit; 


END; 

的基本思路是摆脱光标并让Oracle运行设置操作（设置为如维恩图，而不是设置为设置东西=）。它会做比RAT（一次行）快得多的速度。

我不是100％肯定的语法，但它从here，以及大约三分之一下来你有这样的页面：

SET（列名，列名，...）=（subquery4）

将按子查询4从数据库中检索到的值分配给column_name列表中的列。子查询必须返回一行，其中包含列出的所有列。子查询返回的列值按顺序分配给列表列中的列。第一个值分配给列表中的第一个列，第二个值分配给列表中的第二个列，依此类推。