Oracle-多列统计信息

这个是对动态采样的一个补充。

人为的建立一个组合列统计信息，只要谓词条件中出现组合列某一列的时候，就不在使用以前的计算方式，从而避免CBO执行计划评估误差问题。

建立组合列：

dbms_stats.create_extended_stats('SH','T2','(n1,n2)');

收集组合列统计信息：

SQL> exec dbms_stats.gather_table_stats('SH','T2',method_opt=>'for columns (n1,n2) size auto');

 

模拟环境：

仍是拿原先的环境试验；

 

 

SQL> select count(1) from t2 where n1=3 and n2=3 and c1='a';

 

COUNT(1)

----------

498

返回结果集为498

SQL> select * from table(dbms_xplan.display_cursor(null,0));

 

PLAN_TABLE_OUTPUT

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

SQL_ID 1tpvtscuhy3b6, child number 1

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

select count(1) from t2 where n1=3 and n2=3 and c1='a'

 

Plan hash value: 4191549303

 

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

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

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

| 0 | SELECT STATEMENT | | | | 3 (100)| |

| 1 | SORT AGGREGATE | | 1 | 8 | | |

|* 2 | INDEX RANGE SCAN| IDX_T2 | 25 | 200 | 3 (0)| 00:00:01 |

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

 

Predicate Information (identified by operation id):

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

 

2 - access("N1"=3 AND "N2"=3 AND "C1"='a')

 

Note

-----

- cardinality feedback used for this statement

 

查看执行计划，发现执行计划返回的rows为 25，跟实际返回结果集498严重误差。

 

建立组合列：

SQL> set serveroutput on

SQL> declare

cg_name varchar2(30);

begin

cg_name:=sys.dbms_stats.create_extended_stats('SH','T2','(n1,n2)');

end;

/

 

 

SQL> select object_id from user_objects where object_name='T2';

 

OBJECT_ID

----------

87688

 

SQL> select col#,intcol#,name from sys.col$ where obj#=87688;

 

COL# INTCOL# NAME

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

1 1 C1

2 2 C2

3 3 N1

4 4 N2

0 5 SYS_STUBZH0IHA7K$KEBJVXO5LOHAS

 

能够看到，建立组合列实际上就是在原表T2上附件了一个‘SYS_STUBZH0IHA7K$KEBJVXO5LOHAS

’列。

 

SQL> exec dbms_stats.gather_table_stats('SH','T2',method_opt=>'for columns (n1,n2) size auto');

 

SQL> select count(1) from t2 where n1=3 and n2=3 and c1='a';

 

COUNT(1)

----------

498

SQL> /

 

COUNT(1)

----------

498

 

必定要连续执行2次，

SQL> select * from table(dbms_xplan.display_cursor(null,0));

 

PLAN_TABLE_OUTPUT

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

SQL_ID 1ngdx15skn5cm, child number 1

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

select count(1) from t2 where n1=3 and n2=3 and c1='a'

 

Plan hash value: 4191549303

 

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

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

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

| 0 | SELECT STATEMENT | | | | 3 (100)| |

| 1 | SORT AGGREGATE | | 1 | 8 | | |

|* 2 | INDEX RANGE SCAN| IDX_T2 | 498 | 3984 | 3 (0)| 00:00:01 |

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

 

Predicate Information (identified by operation id):

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

 

2 - access("N1"=3 AND "N2"=3 AND "C1"='a')

 

Note

-----

- cardinality feedback used for this statement

 

查看执行计划，发现返回的rows 是 498 跟实际返回结果集一致。

得出结论：经过建立多列统计信息，也能够实现动态采样的功能。