温馨提示×

温馨提示×

您好,登录后才能下订单哦!

密码登录×
登录注册×
其他方式登录
点击 登录注册 即表示同意《亿速云用户服务条款》

oracle中如何改写exists降低逻辑读

发布时间:2021-11-10 14:05:25 来源:亿速云 阅读:132 作者:小新 栏目:关系型数据库

这篇文章主要介绍oracle中如何改写exists降低逻辑读,文中介绍的非常详细,具有一定的参考价值,感兴趣的小伙伴们一定要看完!

通过将exists改写成in或这inner join优化sql。

Sql_id:056bs9dzz8mwy

问题简述:逻辑读高。

Sql文本:

SELECT A.*, a.rowid

    FROM WBANK.WD_BANK_BASEINFOMATION A

 WHERE EXISTS (SELECT 1

            FROM (select KEYWORD, TYPECODE, INNERCODE, COUNT(*)

                  FROM   WBANK.WD_BANK_BASEINFOMATION

                 WHERE SUBSTR(TYPECODE, 1, 3)   = '001'

                 GROUP BY KEYWORD, TYPECODE,   INNERCODE

                HAVING COUNT(*) <> 1) B

           WHERE A.KEYWORD = B.KEYWORD

             AND A.TYPECODE = B.TYPECODE

             AND A.INNERCODE = B.INNERCODE);

执行计划:

Execution Plan

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

Plan hash value: 1318914978

 

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

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

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

|     0 | SELECT STATEMENT         |                        |     1 |     130 |  7930K  (1)| 39:39:10 |

|*    1 |  FILTER                |                        |       |         |            |          |

|     2 |   TABLE ACCESS FULL    | WD_BANK_BASEINFOMATION |  2640K|     327M|  6249   (2)| 00:01:53 |

|*    3 |   FILTER               |                        |       |       |            |          |

|     4 |    SORT GROUP BY   NOSORT|                        |     1 |      47 |     3   (0)| 00:00:01 |

|*    5 |     INDEX RANGE SCAN   | IDX_WD_B_BI            |     1 |      47 |     3   (0)| 00:00:01 |

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

 

Predicate Information (identified by   operation id):

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

 

     1 - filter( EXISTS (SELECT 0 FROM   "WBANK"."WD_BANK_BASEINFOMATION"

                "WD_BANK_BASEINFOMATION" WHERE   "TYPECODE"=:B1 AND "KEYWORD"=:B2 AND   "INNERCODE"=:B3 AND

                SUBSTR("TYPECODE",1,3)='001' GROUP BY   "KEYWORD","TYPECODE","INNERCODE" HAVING

              COUNT(*)<>1))

     3 - filter(COUNT(*)<>1)

     5 - access("KEYWORD"=:B1 AND "TYPECODE"=:B2 AND   "INNERCODE"=:B3)

         filter("INNERCODE"=:B1 AND   SUBSTR("TYPECODE",1,3)='001')

 

 

Statistics

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

            1  recursive calls

            0  db block gets

      2329554    consistent gets

           13  physical reads

            0  redo size

         2507  bytes sent via SQL*Net to   client

          513  bytes received via SQL*Net   from client

            1  SQL*Net roundtrips to/from   client

            0  sorts (memory)

            0  sorts (disk)

            0  rows processed

可以发现逻辑读高达200多万。

刚看到这个sql的时候猜想会不会逻辑有问题,导致结果集为空。跑了一遍发现结果集确实为空。子查询的innercode列全部为null。根据条件A.INNERCODE = B.INNERCODE外部表(虽然是同一张表)是不会有匹配结果的。转念一想如果子查询innercode列有非空的,那就不会有问题了。当然了还是要询问开发结果集与该列为空是否有必然联系,如果有联系的话可以利用该逻辑关系改写sql。当然,这是后话了。

看一下数据分布:

SQL> select count(*) from WBANK.WD_BANK_BASEINFOMATION;

 

    COUNT(*)

----------

     2645546

 

SQL> select count(*) from (select   KEYWORD, TYPECODE, INNERCODE, COUNT(*)

    2                    FROM WBANK.WD_BANK_BASEINFOMATION

    3                   WHERE   SUBSTR(TYPECODE, 1, 3) = '001'

    4                   GROUP BY   KEYWORD, TYPECODE, INNERCODE

    5                  HAVING   COUNT(*) <> 1);

 

    COUNT(*)

----------

         128

外层结果集是全表数据260多万。子查询结果集只有128条。而根据oracle对exists的处理,会以外部结果集为驱动,也就是说要执行260多万次,这显然是不合理的。如果外部结果集大,内部结果集小的话,这种情况下通常是要用in,以内部结果集为驱动,这样也就执行128次。

验证一下执行次数的问题:

SQL> alter session set   statistics_level=all;

SQL> SELECT A.*, a.rowid

    2    FROM   WBANK.WD_BANK_BASEINFOMATION A

    3   WHERE EXISTS (SELECT 1

    4            FROM (select   KEYWORD, TYPECODE, INNERCODE, COUNT(*)

    5                    FROM   WBANK.WD_BANK_BASEINFOMATION

    6                   WHERE   SUBSTR(TYPECODE, 1, 3) = '001'

    7                   GROUP BY   KEYWORD, TYPECODE, INNERCODE

    8                  HAVING   COUNT(*) <> 1) B

    9           WHERE A.KEYWORD =   B.KEYWORD

 10             AND A.TYPECODE = B.TYPECODE

 11             AND A.INNERCODE = B.INNERCODE);

no rows selected

 

SQL> SELECT * FROM   TABLE(dbms_xplan.display_cursor(null,null,'ALLSTATS LAST'));

 

 

 

Plan hash value: 1318914978

 

PLAN_TABLE_OUTPUT

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

 

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

| Id    | Operation              |   Name                   | Starts | E-Rows | A-Rows |   A-Time     | Buffers |

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

|     0 | SELECT STATEMENT         |                        |      1 |          |      0 |00:00:09.75 |    2329K|

|*    1 |  FILTER                |                        |      1 |        |        0 |00:00:09.75 |    2329K|

|     2 |   TABLE ACCESS FULL    | WD_BANK_BASEINFOMATION |      1 |     2640K|   2645K|00:00:00.61   |   12226 |

|*    3 |   FILTER               |                        |   2632K|        |        0 |00:00:07.38 |    2317K|

|     4 |    SORT GROUP BY   NOSORT|                        |   2632K|      1 |     1273K|00:00:06.64 |    2317K|

|*    5 |     INDEX RANGE SCAN   | IDX_WD_B_BI            |   2632K|      1 |     1273K|00:00:03.42 |    2317K|

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

 

PLAN_TABLE_OUTPUT

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

 

Predicate Information (identified by   operation id):

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

 

     1 - filter( IS NOT NULL)

   3 - filter(COUNT(*)<>1)

     5 - access("KEYWORD"=:B1 AND "TYPECODE"=:B2 AND   "INNERCODE"=:B3)

         filter(("INNERCODE"=:B1 AND   SUBSTR("TYPECODE",1,3)='001'))

 

 

31 rows selected.

可以看到starts列部分,内部子查询2632k次,与外表数据量吻合。

用in改写sql

SELECT A.*, a.rowid

    FROM WBANK.WD_BANK_BASEINFOMATION A

 WHERE (A.KEYWORD,A.TYPECODE,A.INNERCODE) in   (SELECT B.KEYWORD,B.TYPECODE,B.INNERCODE

            FROM (select KEYWORD, TYPECODE, INNERCODE, COUNT(*)

                  FROM   WBANK.WD_BANK_BASEINFOMATION

                 WHERE SUBSTR(TYPECODE, 1, 3)   = '001'

                 GROUP BY KEYWORD, TYPECODE,   INNERCODE

                HAVING COUNT(*) <> 1) B

); 

执行计划:

Set autotrace on

执行sql。

 

得到执行计划:

Execution Plan

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

Plan hash value: 1385212545

 

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

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

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

|     0 | SELECT STATEMENT               |                        |     3 |    7008 |  6236   (2)| 00:01:53 |

|     1 |  NESTED LOOPS                |                        |     3 |    7008 |  6236   (2)| 00:01:53 |

|     2 |   NESTED LOOPS               |                        |     3 |    7008 |  6236   (2)| 00:01:53 |

|     3 |    VIEW                      | VW_NSO_1               |    55 |     118K|  6228   (2)| 00:01:53 |

|*    4 |     FILTER                   |                        |         |       |            |          |

|     5 |      HASH GROUP BY           |                        |     1 |    2585 |  6228   (2)| 00:01:53 |

|*    6 |       TABLE ACCESS FULL      | WD_BANK_BASEINFOMATION | 26410 |  1212K|    6226   (2)| 00:01:53 |

|*    7 |    INDEX RANGE SCAN          | IDX_WD_B_BI            |     1 |         |     2   (0)| 00:00:01 |

|     8 |   TABLE ACCESS BY INDEX   ROWID| WD_BANK_BASEINFOMATION |     1   |   130 |     3     (0)| 00:00:01 |

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

 

Predicate Information (identified by   operation id):

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

 

     4 - filter(COUNT(*)<>1)

     6 - filter(SUBSTR("TYPECODE",1,3)='001')

     7 - access("A"."KEYWORD"="KEYWORD" AND   "A"."TYPECODE"="TYPECODE" AND

                "A"."INNERCODE"="INNERCODE")

         filter("A"."INNERCODE" IS NOT NULL AND   "A"."INNERCODE"="INNERCODE")

 

 

Statistics

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

            1  recursive calls

            0  db block gets

      12226  consistent gets

            0  physical reads

            0  redo size

         2507  bytes sent via SQL*Net to   client

          513  bytes received via SQL*Net   from client

            1  SQL*Net roundtrips to/from   client

            0  sorts (memory)

            0  sorts (disk)

            0  rows processed

执行计划已经变成以内部子查询为驱动表了。而且逻辑读从200万降低1万。

下面再来验证执行次数:

SQL> alter session set   statistics_level=all;

执行sql

SQL> SELECT * FROM   TABLE(dbms_xplan.display_cursor(null,null,'ALLSTATS LAST'));

得到执行计划(部分):

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

--

| Id    | Operation             |   Name                   | Starts |   E-Rows | A-Rows |   A-Time   | Buffers |  OMem |

 |

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

--

|     0 | SELECT STATEMENT      |                        |      1 |        |        0 |00:00:02.28 |   12226 |       |

 |

|*    1 |  HASH JOIN RIGHT SEMI |                        |      1 |     1311K|      0 |00:00:02.28   |   12226 |   391K|

)|

|     2 |   VIEW                | VW_NSO_1               |      1 |    80389 |    128 |00:00:02.28   |   12226 |       |

 |

|*    3 |    FILTER             |                        |      1 |        |      128 |00:00:02.28 |   12226   |       |

 |

|     4 |     HASH GROUP BY     |                        |      1 |     4020 |   1607K|00:00:02.17   |   12226 |   710M|

)|

|*    5 |      TABLE ACCESS FULL|   WD_BANK_BASEINFOMATION |      1 |   1607K|     1607K|00:00:00.78 |   12226   |       |

 |

|*    6 |   TABLE ACCESS FULL   | WD_BANK_BASEINFOMATION |      0 |     1311K|      0 |00:00:00.01   |       0 |       |

 |

发现执行计划并不一致,这个才是真正的执行计划。

 

Predicate Information (identified by   operation id):

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

 

     1 - access("A"."KEYWORD"="KEYWORD" AND   "A"."TYPECODE"="TYPECODE" AND   "A"."INNERCODE"="INNERCODE")

     3 - filter(COUNT(*)<>1)

     5 - filter(SUBSTR("TYPECODE",1,3)='001')

     6 - filter("A"."INNERCODE" IS NOT NULL)

 

Note

 

PLAN_TABLE_OUTPUT

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

-----

     - cardinality feedback used for this statement

后面发现了基数反馈的东西。估计值是实际值差别还是很大的。说明统计信息是有问题的。

查看统计信息已经是4月份收集的了。

收集统计信息

SQL> exec   dbms_stats.gather_table_stats(ownname => 'WBANK',tabname =>   'WD_BANK_BASEINFOMATION',estimate_percent => 10,method_opt=> 'for all   columns size repeat',no_invalidate=>false);

 

PL/SQL procedure successfully completed.

收集完统计信息后的执行计划

Plan hash value: 1385212545

 

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

| Id    | Operation                    |   Name                   | Starts |   E-Rows | A-Rows |   A-Time   | Buffers |  OMem |    1Mem | Used-Mem |

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

|     0 | SELECT STATEMENT               |                        |      1 |        |        2 |00:00:02.55 |   12232 |       |         |    |

|     1 |  NESTED LOOPS                |                        |      1 |        3 |      2 |00:00:02.55 |   12232 |       |         |    |

|     2 |   NESTED LOOPS               |                        |      1 |        3 |      2 |00:00:02.55 |   12230 |       |         |    |

|     3 |    VIEW                      | VW_NSO_1               |      1 |       53 |    129 |00:00:02.55 |   12226 |       |         |    |

|*    4 |     FILTER                   |                        |      1 |        |      129 |00:00:02.55 |   12226   |       |       |      |

|     5 |      HASH GROUP BY           |                        |      1   |      1 |   1607K|00:00:02.40 |   12226 |     710M|    17M|  170M (0)|

|*    6 |       TABLE ACCESS FULL      | WD_BANK_BASEINFOMATION |      1 |    26458 |   1607K|00:00:00.80   |   12226 |       |         |    |

|*    7 |    INDEX RANGE SCAN          | IDX_WD_B_BI            |    129 |      1 |        2 |00:00:00.01 |       4 |       |         |    |

|     8 |   TABLE ACCESS BY INDEX   ROWID| WD_BANK_BASEINFOMATION |      2   |      1 |      2 |00:00:00.01 |       2 |       |         |    |

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

 

Predicate Information (identified by   operation id):

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

 

     4 - filter(COUNT(*)<>1)

     6 - filter(SUBSTR("TYPECODE",1,3)='001')

     7 - access("A"."KEYWORD"="KEYWORD" AND   "A"."TYPECODE"="TYPECODE" AND   "A"."INNERCODE"="INNERCODE")

         filter(("A"."INNERCODE" IS NOT NULL AND   "A"."INNERCODE"="INNERCODE"))

 

 

34 rows selected.

可以看到确实是129次。而且也不存在基数反馈导致执行计划改变了。逻辑读还是在1万多。

突然想到还可以使用inner join的方法来改写sql

SELECT A.*, a.rowid

    FROM WBANK.WD_BANK_BASEINFOMATION A

inner join (select KEYWORD, TYPECODE,   INNERCODE, COUNT(*)

                  FROM   WBANK.WD_BANK_BASEINFOMATION

                 WHERE SUBSTR(TYPECODE, 1, 3)   = '001'

                 GROUP BY KEYWORD, TYPECODE,   INNERCODE

                HAVING COUNT(*) <> 1) B

          on A.KEYWORD = B.KEYWORD

           AND A.TYPECODE = B.TYPECODE

             AND A.INNERCODE = B.INNERCODE;

执行计划:

Plan hash value: 4254729379

 

PLAN_TABLE_OUTPUT

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

 

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

| Id    | Operation                    |   Name                   | Starts |   E-Rows | A-Rows |   A-Time   | Buffers |  OMem |    1Mem | Used-Mem |

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

|     0 | SELECT STATEMENT               |                        |      1 |        |        2 |00:00:02.48 |   12232 |       |         |    |

|     1 |  NESTED LOOPS                |                        |      1 |       59 |      2 |00:00:02.48 |   12232 |       |         |    |

|     2 |   NESTED LOOPS               |                        |      1 |       59 |      2 |00:00:02.48 |   12230 |       |         |    |

|     3 |    VIEW                      |                        |      1 |       59 |    129 |00:00:02.48 |   12226 |       |         |    |

|*    4 |     FILTER                   |                        |      1 |        |      129 |00:00:02.48 |   12226   |       |       |      |

|     5 |      HASH GROUP BY           |                        |      1 |       59 |   1607K|00:00:02.31 |   12226 |     710M|    17M|  168M (0)|

|*    6 |       TABLE ACCESS FULL      | WD_BANK_BASEINFOMATION |      1   |  26466 |   1607K|00:00:00.76 |   12226 |       |         |    |

 

PLAN_TABLE_OUTPUT

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

|*    7 |    INDEX RANGE SCAN          | IDX_WD_B_BI            |    129 |      1 |        2 |00:00:00.01 |       4 |       |         |    |

|     8 |   TABLE ACCESS BY INDEX   ROWID| WD_BANK_BASEINFOMATION |      2   |      1 |      2 |00:00:00.01 |       2 |       |         |    |

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

 

Predicate Information (identified by   operation id):

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

 

     4 - filter(COUNT(*)<>1)

     6 - filter(SUBSTR("TYPECODE",1,3)='001')

     7 -   access("A"."KEYWORD"="B"."KEYWORD"   AND "A"."TYPECODE"="B"."TYPECODE" AND   "A"."INNERCODE"="B"."INNERCODE")

         filter(("A"."INNERCODE" IS NOT NULL AND   "A"."INNERCODE"="B"."INNERCODE"))

 

PLAN_TABLE_OUTPUT

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

 

 

34 rows selected.

逻辑读

 

Statistics

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

            1  recursive calls

            0  db block gets

        12232    consistent gets

            0  physical reads

            0  redo size

         3083  bytes sent via SQL*Net to   client

          524  bytes received via SQL*Net   from client

            2  SQL*Net roundtrips to/from   client

            0  sorts (memory)

            0  sorts (disk)

            2  rows processed

逻辑读也是1万多。

看一下执行计划发现,瓶颈都在对表的全表扫且过滤条件filter(SUBSTR("TYPECODE",1,3)='001')。

可以考虑在这列上建函数索引,

 

SQL> select count(*) from wbank.WD_BANK_BASEINFOMATION   WHERE SUBSTR(TYPECODE, 1, 3) = '001';

 

    COUNT(*)

----------

     1607674

表的数据一共只有2645546,返回1607674,所以建了索引也没用,所以不用建索引了。

综上所述。优化建议是更改sql,将exists改成in或者inner join:

SELECT A.*, a.rowid

    FROM WBANK.WD_BANK_BASEINFOMATION A

 WHERE (A.KEYWORD,A.TYPECODE,A.INNERCODE) in   (SELECT B.KEYWORD,B.TYPECODE,B.INNERCODE

            FROM (select KEYWORD, TYPECODE, INNERCODE, COUNT(*)

                  FROM   WBANK.WD_BANK_BASEINFOMATION

                 WHERE SUBSTR(TYPECODE, 1, 3)   = '001'

                 GROUP BY KEYWORD, TYPECODE,   INNERCODE

                HAVING COUNT(*) <> 1) B

);

 

或者

SELECT A.*, a.rowid

    FROM WBANK.WD_BANK_BASEINFOMATION A

inner join (select KEYWORD, TYPECODE,   INNERCODE, COUNT(*)

                  FROM   WBANK.WD_BANK_BASEINFOMATION

                 WHERE SUBSTR(TYPECODE, 1, 3)   = '001'

                 GROUP BY KEYWORD, TYPECODE,   INNERCODE

                HAVING COUNT(*) <> 1) B

          on A.KEYWORD = B.KEYWORD

           AND A.TYPECODE = B.TYPECODE

             AND A.INNERCODE = B.INNERCODE;

 

逻辑读将从200多万将至1万多。

以上是“oracle中如何改写exists降低逻辑读”这篇文章的所有内容,感谢各位的阅读!希望分享的内容对大家有帮助,更多相关知识,欢迎关注亿速云行业资讯频道!

向AI问一下细节

免责声明:本站发布的内容(图片、视频和文字)以原创、转载和分享为主,文章观点不代表本网站立场,如果涉及侵权请联系站长邮箱:is@yisu.com进行举报,并提供相关证据,一经查实,将立刻删除涉嫌侵权内容。

AI