Oracle 利用执行计划来避免排序操作

在oracle中，利用index来避免排序

SQL> CREATE TABLE T_NOSORT (ID NUMBER PRIMARY KEY, NAME VARCHAR2(30) NOT NULL);

SQL> CREATE INDEX IND_T_NOSORT_NAME ON T_NOSORT(NAME);

SQL> INSERT INTO T_NOSORT SELECT ROWNUM, TABLE_NAME FROM USER_TABLES;

SQL> COMMIT;

SQL> SET AUTOT ON EXP

SQL> SELECT ID, NAME FROM T_NOSORT ORDER BY NAME;

----------------------------------------------------------

Plan hash value: 1041838668

-------------------------------------------------------

| Id | Operation | Name | Rows | Bytes |

-------------------------------------------------------

| 0 | SELECT STATEMENT | | 15 | 450 |

| 1 | SORT ORDER BY | | 15 | 450 |

| 2 | TABLE ACCESS FULL| T_NOSORT | 15 | 450 |

-------------------------------------------------------

QL> SELECT /*+ INDEX(T_NOSORT IND_T_NOSORT_NAME) */ ID, NAME FROM T_NOSORT ORDER BY NAME;

| Id  | Operation                   | Name              | Rows  | Bytes | Cost (%CPU)| Time     |

-------------------------------------------------------------------------------|   0 | SELECT STATEMENT          |                   |    41 |  1230 |   827   (1)| 00:00:10 |

|   1 |  TABLE ACCESS BY INDEX ROWID| T_NOSORT          |    41 |  1230 |   827   (1)| 00:00:10 |

|   2 |   INDEX FULL SCAN           | IND_T_NOSORT_NAME |    41 |       |    26   (0)| 00:00:01 |

利用索引范围扫描

SQL> SELECT ID, NAME FROM T_NOSORT WHERE NAME < 'I' ORDER BY NAME;

----------------------------------------------------------

Plan hash value: 1041838668

-------------------------------------------------------

| Id | Operation | Name | Rows | Bytes |

-------------------------------------------------------

| 0 | SELECT STATEMENT | | 9 | 270 |

| 1 | SORT ORDER BY | | 9 | 270 |

|* 2 | TABLE ACCESS FULL| T_NOSORT | 9 | 270 |

-------------------------------------------------------

Predicate Information (identified by operation id):

---------------------------------------------------

2 - filter("NAME"<'I')

SQL> SELECT /*+ INDEX(T_NOSORT IND_T_NOSORT_NAME) */ ID, NAME 

2 FROM T_NOSORT WHERE NAME < 'I' ORDER BY NAME;

----------------------------------------------------------

Plan hash value: 919790285

-------------------------------------------------------------------------

| Id | Operation | Name | Rows | Bytes |

-------------------------------------------------------------------------

| 0 | SELECT STATEMENT | | 9 | 270 |

| 1 | TABLE ACCESS BY INDEX ROWID| T_NOSORT | 9 | 270 |

|* 2 | INDEX RANGE SCAN | IND_T_NOSORT_NAME | 9 | |

-------------------------------------------------------------------------

Predicate Information (identified by operation id):

---------------------------------------------------

2 - access("NAME"<'I')

filter("NAME"<'I')

如果是倒序排序

QL> SELECT /*+ INDEX_DESC(T_NOSORT IND_T_NOSORT_NAME) */ ID, NAME 

2 FROM T_NOSORT ORDER BY NAME DESC;

----------------------------------------------------------

Plan hash value: 2858378269

-------------------------------------------------------------------------

| Id | Operation | Name | Rows | Bytes |

-------------------------------------------------------------------------

| 0 | SELECT STATEMENT | | 15 | 450 |

| 1 | TABLE ACCESS BY INDEX ROWID| T_NOSORT | 15 | 450 |

| 2 | INDEX FULL SCAN DESCENDING| IND_T_NOSORT_NAME | 15 | |

-------------------------------------------------------------------------

这里只是说明了能够避免排序的执行计划，但是不一定能提升性能

上面针对于单表

MERGE JOIN连接方式

SQL> CREATE TABLE T1 (ID NUMBER PRIMARY KEY, NAME VARCHAR2(30) NOT NULL);

SQL> CREATE TABLE T2 (ID NUMBER, NAME VARCHAR2(30));

SQL> CREATE INDEX IND_T1_NAME ON T1(NAME);

SQL> INSERT INTO T1 SELECT ROWNUM, TABLE_NAME FROM USER_TABLES;

SQL> INSERT INTO T2 SELECT ROWNUM, OBJECT_NAME FROM USER_OBJECTS;

SQL> COMMIT;

SQL> SET AUTOT ON EXP

SQL> SELECT /*+ USE_MERGE(T1, T2) */ T2.NAME, T1.ID FROM T1, T2

 WHERE T1.NAME = T2.NAME ;

| Id  | Operation           | Name | Rows  | Bytes | Cost (%CPU)| Time     |

----------------------------------------------------------------------------

|   0 | SELECT STATEMENT    |      |    41 |  1927 |     8  (25)| 00:00:01 |

|   1 |  MERGE JOIN         |      |    41 |  1927 |     8  (25)| 00:00:01 |

|   2 |   SORT JOIN         |      |    41 |  1230 |     4  (25)| 00:00:01 |

|   3 |    TABLE ACCESS FULL| T1   |    41 |  1230 |     3   (0)| 00:00:01 |

|*  4 |   SORT JOIN         |      |    99 |  1683 |     4  (25)| 00:00:01 |

|   5 |    TABLE ACCESS FULL| T2   |    99 |  1683 |     3   (0)| 00:00:01 |

----------------------------------------------------

Predicate Information (identified by operation id):

---------------------------------------------------

4 - access("T1"."NAME"="T2"."NAME")

filter("T1"."NAME"="T2"."NAME")

SQL> SELECT /*+ USE_MERGE(T2, T1) */ T2.NAME, T1.ID

 FROM T1, T2 WHERE T1.NAME = T2.NAME ORDER BY T2.NAME ;

----------------------------------------------------

Plan hash value: 412793182

----------------------------------------------------

| Id | Operation | Name | Rows | Bytes |

----------------------------------------------------

| 0 | SELECT STATEMENT | | 19 | 893 |

| 1 | MERGE JOIN | | 19 | 893 |

| 2 | SORT JOIN | | 17 | 510 |

| 3 | TABLE ACCESS FULL| T1 | 17 | 510 |

|* 4 | SORT JOIN | | 97 | 1649 |

| 5 | TABLE ACCESS FULL| T2 | 97 | 1649 |

----------------------------------------------------

Predicate Information (identified by operation id):

---------------------------------------------------

4 - access("T1"."NAME"="T2"."NAME")

filter("T1"."NAME"="T2"."NAME")

针对merge join只能对连接的列排序，且排序操作只能是升序

SQL> SELECT /*+ USE_MERGE(T2, T1) */ T2.NAME, T1.ID

 FROM T1, T2 WHERE T1.NAME = T2.NAME ORDER BY T2.NAME DESC;

|   0 | SELECT STATEMENT     |      |    41 |  1927 |     9  (34)| 00:00:01 |

|   1 |  SORT ORDER BY       |      |    41 |  1927 |     9  (34)| 00:00:01 |

|   2 |   MERGE JOIN         |      |    41 |  1927 |     8  (25)| 00:00:01 |

|   3 |    SORT JOIN         |      |    41 |  1230 |     4  (25)| 00:00:01 |

|   4 |     TABLE ACCESS FULL| T1   |    41 |  1230 |     3   (0)| 00:00:01 |

|*  5 |    SORT JOIN         |      |    99 |  1683 |     4  (25)| 00:00:01 |

|   6 |     TABLE ACCESS FULL| T2   |    99 |  1683 |     3   (0)| 00:00:01 |5 - access("T1"."NAME"="T2"."NAME")

filter("T1"."NAME"="T2"."NAME")

NESTED LOOP连接

由于nested loop不会对操作结果排序，所以结果是无序的

如果驱动表在连接前是有序的，在连接后，结果还是有序的

SQL> CREATE INDEX IND_T2_ID ON T2(ID);

SQL> CREATE INDEX IND_T2_NAME ON T2(NAME);

SQL> SELECT /*+ USE_NL(T1, T2) */ T1.ID, T1.NAME, T2.NAME 

2 FROM T1, T2

3 WHERE T1.ID = T2.ID

4 ;

----------------------------------------------------------

Plan hash value: 3621112097

----------------------------------------------------------------------

| Id | Operation | Name | Rows | Bytes | Cost

----------------------------------------------------------------------

| 0 | SELECT STATEMENT | | 17 | 1020 | 

| 1 | TABLE ACCESS BY INDEX ROWID| T2 | 1 | 30 | 

| 2 | NESTED LOOPS | | 17 | 1020 | 

| 3 | TABLE ACCESS FULL | T1 | 17 | 510 | 

|* 4 | INDEX RANGE SCAN | IND_T2_ID | 1 | | 

----------------------------------------------------------------------

Predicate Information (identified by operation id):

---------------------------------------------------

4 - access("T1"."ID"="T2"."ID")

SQL> SELECT /*+ ORDERED INDEX(T1 IND_T1_NAME) USE_NL(T1, T2) */ 

2 T1.ID, T1.NAME, T2.NAME 

3 FROM T1, T2

4 WHERE T1.ID = T2.ID

5 ORDER BY T1.NAME

6 ;

----------------------------------------------------------

Plan hash value: 1062594094

---------------------------------------------------------------------

| Id | Operation | Name | Rows | Bytes |

---------------------------------------------------------------------

| 0 | SELECT STATEMENT | | 17 | 1020 |

| 1 | TABLE ACCESS BY INDEX ROWID | T2 | 1 | 30 |

| 2 | NESTED LOOPS | | 17 | 1020 |

| 3 | TABLE ACCESS BY INDEX ROWID| T1 | 17 | 510 |

| 4 | INDEX FULL SCAN | IND_T1_NAME | 17 | |

|* 5 | INDEX RANGE SCAN | IND_T2_ID | 1 | |

---------------------------------------------------------------------

Predicate Information (identified by operation id):

---------------------------------------------------

5 - access("T1"."ID"="T2"."ID")

SQL> ALTER TABLE T2 MODIFY NAME NOT NULL;

SQL> CREATE INDEX IND_T2_NAME ON T2(NAME);

SQL> SET AUTOT OFF

SQL> UPDATE T2 SET ID = MOD(ID, 17) + 1;

SQL> COMMIT;

SQL> SET AUTOT ON EXP

SQL> SELECT /*+ ORDERED INDEX(T1 IND_T1_NAME) 

2 INDEX(T2 IND_T2_NAME) USE_NL(T1, T2) */ 

3 T1.ID, T1.NAME, T2.NAME 

4 FROM T1, T2

5 WHERE T1.ID = T2.ID

6 ORDER BY T1.NAME

7 ;

----------------------------------------------------------

Plan hash value: 3719138605

---------------------------------------------------------------------

| Id | Operation | Name | Rows | Bytes |

---------------------------------------------------------------------

| 0 | SELECT STATEMENT | | 97 | 5820 |

|* 1 | TABLE ACCESS BY INDEX ROWID | T2 | 6 | 180 |

| 2 | NESTED LOOPS | | 97 | 5820 |

| 3 | TABLE ACCESS BY INDEX ROWID| T1 | 17 | 510 |

| 4 | INDEX FULL SCAN | IND_T1_NAME | 17 | |

| 5 | INDEX FULL SCAN | IND_T2_NAME | 97 | |

---------------------------------------------------------------------

Predicate Information (identified by operation id):

---------------------------------------------------

1 - filter("T1"."ID"="T2"."ID")

SQL> SELECT /*+ ORDERED INDEX(T1 IND_T1_NAME) 

2 INDEX_DESC(T2 IND_T2_NAME) USE_NL(T1, T2) */ 

3 T1.ID, T1.NAME, T2.NAME 

4 FROM T1, T2

5 WHERE T1.ID = T2.ID

6 ORDER BY T1.NAME

7 ;

----------------------------------------------------------

Plan hash value: 2531946081

---------------------------------------------------------------------

| Id | Operation | Name | Rows | Bytes |

---------------------------------------------------------------------

| 0 | SELECT STATEMENT | | 97 | 5820 |

|* 1 | TABLE ACCESS BY INDEX ROWID | T2 | 6 | 180 |

| 2 | NESTED LOOPS | | 97 | 5820 |

| 3 | TABLE ACCESS BY INDEX ROWID| T1 | 17 | 510 |

| 4 | INDEX FULL SCAN | IND_T1_NAME | 17 | |

| 5 | INDEX FULL SCAN DESCENDING | IND_T2_NAME | 97 | |

---------------------------------------------------------------------

Predicate Information (identified by operation id):

---------------------------------------------------

1 - filter("T1"."ID"="T2"."ID")

SQL> SELECT /*+ ORDERED INDEX(T1 IND_T1_NAME) 

2 INDEX(T2 IND_T2_NAME) USE_NL(T1, T2) */ 

3 T1.ID, T1.NAME, T2.NAME 

4 FROM T1, T2

5 WHERE T1.ID = T2.ID

6 ORDER BY T1.NAME, T2.NAME DESC

7 ;

----------------------------------------------------------

Plan hash value: 1438746903

----------------------------------------------------------------------

| Id | Operation | Name | Rows | Bytes |

----------------------------------------------------------------------

| 0 | SELECT STATEMENT | | 97 | 5820 |

| 1 | SORT ORDER BY | | 97 | 5820 |

|* 2 | TABLE ACCESS BY INDEX ROWID | T2 | 6 | 180 |

| 3 | NESTED LOOPS | | 97 | 5820 |

| 4 | TABLE ACCESS BY INDEX ROWID| T1 | 17 | 510 |

| 5 | INDEX FULL SCAN | IND_T1_NAME | 17 | |

| 6 | INDEX FULL SCAN | IND_T2_NAME | 97 | |

----------------------------------------------------------------------

Predicate Information (identified by operation id):

---------------------------------------------------

2 - filter("T1"."ID"="T2"."ID")

Oracle 利用执行计划来避免排序操作的更多相关文章

Oracle sql执行计划解析
Oracle sql执行计划解析 https://blog.csdn.net/xybelieve1990/article/details/50562963 Oracle优化器 Oracle的优化器共有 ...
怎样看懂Oracle的执行计划
怎样看懂Oracle的执行计划一.什么是执行计划 An explain plan is a representation of the access path that is taken when ...
（转）Oracle定时执行计划任务
Oracle定时执行计划任务在日常工作中,往往有些事情是需要经常重复地做的,例如每天更新业务报表.每天从数据库中提取符合条件的数据.每天将客户关系管理系统中的数据分配给员工做数据库营销……因此我们就 ...
【ORACLE】记录通过执行Oracle的执行计划查询SQL脚本中的效率问题
记录通过执行Oracle的执行计划查询SQL脚本中的效率问题问题现象: STARiBOSS5.8.1R2版本中,河北对帐JOB执行时,无法生成发票对帐文件. 首先,Quartz表达式培植的启 ...
Oracle的执行计划（来自百度文库）
如何开启oracle执行计划 http://wenku.baidu.com/view/7d1ff6bc960590c69ec37636.html怎样看懂Oracle的执行计划 http://wenku ...
ORACLE的执行计划
转自:http://www.cnblogs.com/lovingprince/archive/2007/12/07/2166400.html 背景知识: 为了更好的进行下面的内容我们必须 ...
oracle 理解执行计划
·BUFFER SORT是BUFFER却不是SORT 用AUTOTRACE查看执行的计划的同学常问到执行计划里的BUFFER SORT是什么意思,这里为什么要排序呢? BUFFER SORT不是一种排 ...
分析oracle的执行计划（explain plan）并对对sql进行优化实践
基于oracle的应用系统很多性能问题,是由应用系统sql性能低劣引起的,所以,sql的性能优化很重要,分析与优化sql的性能我们一般通过查看该sql的执行计划,本文就如何看懂执行计划,以及如何通过分 ...
oracle sql 执行计划分析
转自http://itindex.net/detail/45962-oracle-sql-%E8%AE%A1%E5%88%92 一.首先创建表 SQL> show user USER is &q ...

随机推荐

使用springmvc的时候报错NoSuchBeanDefinitionException: No qualifying bean of type
NoSuchBeanDefinitionException: No qualifying bean of type 其实我至今都不知道错误的根源在哪里,<context:component-sc ...
图片qq浏览器不显示，微信显示问题原因
1.qq浏览器关闭云加速就可以了
Mac 终端 Termial 高亮配置（PS1变量配置）
操作环境: 系统:Mac 10.12 编辑器:vim 一.无脑配置: 1. 打开中端输入: vi ~/.bash_profile 2. 打开并编辑 .bash_profile 文件: 按键盘“i”,进 ...
storm 入门介绍（持续更新）
storm的集群表面上看和hadoop的集群非常像.但是在Hadoop上面你运行的是MapReduce的Job, 而在Storm上面你运行的是Topology.它们是非常不一样的 — 一个关键的区别是 ...
后台操作日志,插入数据获取的该数据主键ID为null
代码如下: @Override public void saveTopicResource(TopicResourceModel model, Integer userId) { TopicResou ...
InnoDB存储引擎的B+树索引算法
关于B+树数据结构 ①InnoDB存储引擎支持两种常见的索引. 一种是B+树,一种是哈希. B+树中的B代表的意思不是二叉(binary),而是平衡(balance),因为B+树最早是从平衡二叉树演化 ...
ndk+opencv安装＋各种错误分析（新版安装，编译不需要Cygwin 和Sequoyah了）
鼓捣了两三天,终于成功算跑通了一个简单的程序.下面说说具体的安装: 因为从同学那里拷过来的eclipse 就有adt cdt 的插件.所以这两个就不用再安装了.(需要的话自己安装) 具体说下安装过程: ...
js获取一周的日期范围
function getWeek() { this.nowTime = new Date(); this.init = function() { this.dayInWeek = this.nowTi ...
基于Python实现的四则运算生成程序
Github项目地址:传送门小组成员:黄晓彬(代码实现) 黄钰城(代码审查) 需求: 1. 使用 -n 参数控制生成题目的个数. 2. 使用 -r 参数控制题目中数值(自然数.真分数和真分数分母)的 ...
Java进阶面试问题列表
面向对象编程的基本理念与核心设计思想解释下多态性(polymorphism),封装性(encapsulation),内聚(cohesion)以及耦合(coupling). 继承(Inheritanc ...

Oracle 利用执行计划来避免排序操作

MERGE JOIN连接方式

NESTED LOOP连接

Oracle 利用执行计划来避免排序操作的更多相关文章

随机推荐

热门专题