Oracle SQL Optimizer IN VS Exists Again

最新推荐文章于 2025-09-09 20:40:49 发布

转载最新推荐文章于 2025-09-09 20:40:49 发布 · 60 阅读

0 ·

CC 4.0 BY-SA版权

原文链接：https://yq.aliyun.com/articles/523041

文章标签：

#数据库

本文探讨了在Oracle 11g中，IN与Exists两种SQL写法在现代CBO优化器下的性能表现。通过实验发现，由于CBO优化器会进行查询转换，这两种写法在性能上的差异并不明显。

传统的SQL优化思想认为IN、Exists这2种写法不同数据量的环境中各擅胜场，Developer应当根据实际情况合理运用IN或Exists。实际我认为，这种认识对于现代SQL优化器Optimizer(10g以后)不再适用。 CBO优化器中本身包含了查询转换(Query Transformation)的功能。为什么CBO要做查询转换(Query Transformation)？回答：用户提交给SQL引擎的语句可能不是执行查询最有效的语句查询转换(Query Transformation)做了什么？回答：查询转换按照固定的语法将语句重写为等意的SQL，但是优化了其SQL结构查询转换(Query Transformation)包括启发式地查询转换和基于成本的查询转换。对于IN、Exists , CBO optimizer 的查询装换效果如何？：

SQL> select * from v$version;

BANNER
--------------------------------------------------------------------------------
Oracle Database 11g Enterprise Edition Release 11.2.0.3.0 - 64bit Production
PL/SQL Release 11.2.0.3.0 - Production
CORE    11.2.0.3.0      Production
TNS for 64-bit Windows: Version 11.2.0.3.0 - Production
NLSRTL Version 11.2.0.3.0 - Production

SQL> select count(1) from joinA A where A.CUST_ID in ( select B.cust_id from joinb  B where B.CUST_FIRST_NAME like 'C%');

  COUNT(1)
----------
     39381

SQL> alter system flush shared_pool;

System altered.

SQL> alter system flush shared_pool;

System altered.

SQL> alter session set events '10053 trace name context forever,level 1';

Session altered.

SQL> select count(1) from joinA A where A.CUST_ID in ( select B.cust_id from joinb  B where B.CUST_FIRST_NAME like 'C%');

  COUNT(1)
----------
     39381

SQL> oradebug setmypid
Statement processed.
SQL> oradebug tracefile_name
D:\APP\ML\diag\rdbms\testem\testem\trace\testem_ora_6776.trc

Final query after transformations:******* UNPARSED QUERY IS *******
SELECT COUNT(*) "COUNT(1)"
  FROM (SELECT "B"."CUST_ID" "CUST_ID"
          FROM "SYS"."JOINB" "B"
         WHERE "B"."CUST_FIRST_NAME" LIKE 'C%') "VW_NSO_1",
       "SYS"."JOINA" "A"
 WHERE "A"."CUST_ID" = "VW_NSO_1"."CUST_ID"

kkoqbc: optimizing query block SEL$683B0107 (#2)

SQL> alter session set events '10053 trace name context forever,level 1';

Session altered.

SQL> select count(1) from joinA A where exists ( select 1  from joinb  B where B.CUST_FIRST_NAME like 'C%' and B.cust_id=A.cust_id  );

  COUNT(1)
----------
     39381

Final query after transformations:******* UNPARSED QUERY IS *******

SELECT COUNT(*) "COUNT(1)"
  FROM (SELECT "B"."CUST_ID" "ITEM_1"
          FROM "SYS"."JOINB" "B"
         WHERE "B"."CUST_FIRST_NAME" LIKE 'C%') "VW_SQ_1",
       "SYS"."JOINA" "A"
 WHERE "VW_SQ_1"."ITEM_1" = "A"."CUST_ID"

如以上10053 trace 所显示对于2个分别使用了IN和Exists的查询， CBO Optimizer查询转换后得到的结果十分类似： IN: SELECT COUNT(*) "COUNT(1)" FROM (SELECT "B"."CUST_ID" "CUST_ID" FROM "SYS"."JOINB" "B" WHERE "B"."CUST_FIRST_NAME" LIKE 'C%') "VW_NSO_1", "SYS"."JOINA" "A" WHERE "A"."CUST_ID" = "VW_NSO_1"."CUST_ID" Exists: SELECT COUNT(*) "COUNT(1)" FROM (SELECT "B"."CUST_ID" "ITEM_1" FROM "SYS"."JOINB" "B" WHERE "B"."CUST_FIRST_NAME" LIKE 'C%') "VW_SQ_1", "SYS"."JOINA" "A" WHERE "VW_SQ_1"."ITEM_1" = "A"."CUST_ID" CBO在完成Query transformations之后才会对转换后的语句进行APA(access path analysis)和后续的优化树构建、成本比较等等，一般来说如果你使用10g以后的CBO ，可以认为语义等价的IN或Exists在性能上也不会有太大的区别。

本文转自maclean_007 51CTO博客，原文链接:本文转自maclean_007 51CTO博客，原文链接:http://blog.51cto.com/maclean/1278473