【书评:Oracle查询优化改写】第四章
作者：互联网
BLOG文档结构图
一.1 导读
各位技术爱好者，看完本文后，你可以掌握如下的技能，也可以学到一些其它你所不知道的知识，~O(∩_∩)O~：
① check的特殊用法
② sql优化中使用merge语句代替update语句（重点）
本文如有错误或不完善的地方请大家多多指正，ITPUB留言或QQ皆可，您的批评指正是我写作的最大动力。
一.2 实验环境介绍
目标库：11.2.0.3 RHEL6.5
一.3 前言
前3章的链接参考相关连接：
【书评:Oracle查询优化改写】第一章 http://blog.itpub.net/26736162/viewspace-1652985/
【书评:Oracle查询优化改写】第二章 http://blog.itpub.net/26736162/viewspace-1654252/
【书评:Oracle查询优化改写】第三章 http://blog.itpub.net/26736162/viewspace-1660422/
今天来写写这本书的第四章的内容，第四章主要讲了UPDATE语句的正确用法，以及什么时候UPDATE语句应改写为MERGE， 第四章的内容目录如下：
第 4 章 插入、更新与删除
4.1 插入新记录
4.2 阻止对某几列插入
4.3 复制表的定义及数据
4.4 用 WITH CHECK OPTION 限制数据录入
4.5 多表插入语句
4.6 用其他表中的值更新
4.7 合并记录
4.8 删除违反参照完整性的记录
4.9 删除名称重复的记录
一.4 check的特殊用法
我们知道sysdate不能用于check约束，但是有这种需求的时候怎么办呢？如下例子利用视图加with check option即可解决。
09:39:08 SQL> create table ttt(create_date date check(create_date > sysdate));
create table ttt(create_date date check(create_date > sysdate))
*
ERROR at line 1:
ORA-02436: date or system variable wrongly specified in CHECK constraint
09:41:56 SQL> insert into (select empno,ename,hiredate from scott.emp where hiredate <= sysdate with check option)
09:42:13 2 values ( 9999,'test',sysdate+1);
insert into (select empno,ename,hiredate from scott.emp where hiredate <= sysdate with check option)
*
ERROR at line 1:
ORA-01402: view WITH CHECK OPTION where-clause violation
Elapsed: 00:00:00.12
09:42:14 SQL> insert into (select empno,ename,hiredate from scott.emp where hiredate <= sysdate with check option)
09:42:56 2 values ( 9999,'test',sysdate-1);
1 row created.
Elapsed: 00:00:00.03
09:42:57 SQL>
一.5 merge语句
关于update的一个容易出错的地方就是不写where子句，这样的话会更新掉全表的数据，一个技巧就是把set中的值复制到where子句中即可。
另外，建议大家在做多表关联更新的时候修改为merge语句，因为merge into语句只访问了一次表：
[oracle@rhel6_lhr ~]$ sqlplus / as sysdba
SQL*Plus: Release 11.2.0.3.0 Production on 星期二 5月 19 10:26:55 2015
Copyright (c) 1982, 2011, Oracle. All rights reserved.
连接到:
Oracle Database 11g Enterprise Edition Release 11.2.0.3.0 - 64bit Production
With the Partitioning, Automatic Storage Management, OLAP, Data Mining
and Real Application Testing options
10:26:55 SQL> set autot on;
10:28:05 SQL> alter table lhr.emp_bk add dname varchar2(50) default 'noname';
表已更改。
已用时间: 00: 00: 01.23
10:30:04 SQL> update lhr.emp_bk a
10:30:09 2 set a.dname =(select b.dname from lhr.dept_bk b where b.deptno=a.deptno and b.dname in ('ACCOUNTING','RESERCH'))
10:30:09 3 WHERE EXISTS (select 1 from lhr.dept_bk b where b.deptno=a.deptno and b.dname in ('ACCOUNTING','RESERCH' ))
10:30:09 4 ;
已更新3行。
已用时间: 00: 00: 00.05
执行计划
----------------------------------------------------------
Plan hash value: 3525057516
-------------------------------------------------------------------------------
| Id | Operation | Name | Rows | Bytes | Cost (%CPU)| Time |
-------------------------------------------------------------------------------
| 0 | UPDATE STATEMENT | | 4 | 544 | 28 (18)| 00:00:01 |
| 1 | UPDATE | EMP_BK | | | | |
|* 2 | HASH JOIN SEMI | | 4 | 544 | 8 (13)| 00:00:01 |
| 3 | TABLE ACCESS FULL| EMP_BK | 14 | 1596 | 3 (0)| 00:00:01 |
|* 4 | TABLE ACCESS FULL| DEPT_BK | 1 | 22 | 4 (0)| 00:00:01 |
|* 5 | TABLE ACCESS FULL | DEPT_BK | 1 | 22 | 4 (0)| 00:00:01 |
-------------------------------------------------------------------------------
Predicate Information (identified by operation id):
---------------------------------------------------
2 - access("B"."DEPTNO"="A"."DEPTNO")
4 - filter("B"."DNAME"='ACCOUNTING' OR "B"."DNAME"='RESERCH')
5 - filter("B"."DEPTNO"=:B1 AND ("B"."DNAME"='ACCOUNTING' OR
"B"."DNAME"='RESERCH'))
Note
-----
- dynamic sampling used for this statement (level=2)
统计信息
----------------------------------------------------------
69 recursive calls
13 db block gets
121 consistent gets
9 physical reads
3012 redo size
837 bytes sent via SQL*Net to client
997 bytes received via SQL*Net from client
3 SQL*Net roundtrips to/from client
12 sorts (memory)
0 sorts (disk)
3 rows processed
已用时间: 00: 00: 00.00
10:33:13 SQL> merge into lhr.emp_bk a
10:33:32 2 using (select b.dname,deptno from lhr.dept_bk b where b.dname in ('ACCOUNTING','RESERCH')) bb
10:33:32 3 on (bb.deptno=a.deptno)
10:33:32 4 when matched then
10:33:32 5 update set a.dname =bb.dname
10:33:32 6 ;
3 行已合并。
已用时间: 00: 00: 00.03
执行计划
----------------------------------------------------------
Plan hash value: 1386289611
--------------------------------------------------------------------------------
| Id | Operation | Name | Rows | Bytes | Cost (%CPU)| Time |
--------------------------------------------------------------------------------
| 0 | MERGE STATEMENT | | 4 | 492 | 8 (13)| 00:00:01 |
| 1 | MERGE | EMP_BK | | | | |
| 2 | VIEW | | | | | |
|* 3 | HASH JOIN | | 4 | 592 | 8 (13)| 00:00:01 |
|* 4 | TABLE ACCESS FULL| DEPT_BK | 1 | 22 | 4 (0)| 00:00:01 |
| 5 | TABLE ACCESS FULL| EMP_BK | 14 | 1764 | 3 (0)| 00:00:01 |
--------------------------------------------------------------------------------
Predicate Information (identified by operation id):
---------------------------------------------------
3 - access("DEPTNO"="A"."DEPTNO")
4 - filter("B"."DNAME"='ACCOUNTING' OR "B"."DNAME"='RESERCH')
Note
-----
- dynamic sampling used for this statement (level=2)
统计信息
----------------------------------------------------------
20 recursive calls
7 db block gets
38 consistent gets
1 physical reads
1872 redo size
838 bytes sent via SQL*Net to client
942 bytes received via SQL*Net from client
3 SQL*Net roundtrips to/from client
3 sorts (memory)
0 sorts (disk)
3 rows processed
10:33:32 SQL>
标签：00,01,dname,33,书评,SQL,Oracle,where,第四章 来源： https://blog.51cto.com/lhrbest/2690199