公式：摘自《催华-基于ORACLE的SQL优化》

链接地址：http://blog.csdn.net/launch_225/article/details/25472129

等频：唯一值《=桶数

=>等频(频率)直方图

-->density计算

density=1/(2*num_rows*null_adjust)

null_adjust=(num_rows-num_nulls)/num_rows

-->等值查询,val在low_value and high_value之间,值等于某个endpoint_value

cardinality=num_rows * selectivity

selectivity=bucketsize/num_rows

bucketsize=current_endpoint_number-previous_endpoint_number

TABLE_NAMECOLUMN_NAMENUM_BUCKETSLOW_VALUEHIGH_VALUEHISTOGRAM

1T1N113C102C113FREQUENCY

-->density计算

density=1/(2*num_rows*null_adjust)

SQL> select 1/(2*18*1) from dual;

1/(2*18*1)

----------

.027777778

OWNERTABLE_NAMENUM_DISTINCTDENSITYCOLUMN_NAME

1AIKIT1130.0277777777777778N1

-->等值查询,val在low_value and high_value之间,值等于某个endpoint_value

cardinality=num_rows * selectivity

selectivity=bucketsize/num_rows

bucketsize=current_endpoint_number-previous_endpoint_number

1* select * from t1 where n1=6

SQL> /

Execution Plan

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

Plan hash value: 1577308413

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

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

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

| 0 | SELECT STATEMENT | | 3 | 15 | 2 (0)| 00:00:01 |

| 1 | TABLE ACCESS BY INDEX ROWID| T1 | 3 | 15 | 2 (0)| 00:00:01 |

|* 2 | INDEX RANGE SCAN | IDX_T1_N1 | 3 | | 1 (0)| 00:00:01 |

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

Predicate Information (identified by operation id):

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

2 - access("N1"=6)

OWNERTABLE_NAMECOLUMN_NAMEENDPOINT_NUMBERENDPOINT_VALUE

AIKIT1N111

AIKIT1N132

AIKIT1N144

AIKIT1N155 --previous_endpoint_number

AIKIT1N186 --current_endpoint_number

AIKIT1N199

AIKIT1N11010

AIKIT1N11111

AIKIT1N11312

AIKIT1N11414

AIKIT1N11515

AIKIT1N11716

AIKIT1N11818

bucketsize=8-5=3;

selectivity=3/18

cardinality=18*(3/18)=3;