dbms_stats.gather_table_stats详解

DBMS_STATS.GATHER_TABLE_STATS的语法如下:

DBMS_STATS.GATHER_TABLE_STATS (   ownname          VARCHAR2,

     tabname          VARCHAR2,

     partname         VARCHAR2,

      estimate_percent NUMBER,

     block_sample     BOOLEAN,

      method_opt       VARCHAR2,

      degree          NUMBER,

      granularity      VARCHAR2,

     cascade          BOOLEAN,

      stattab          VARCHAR2,

     statid           VARCHAR2,

      statown          VARCHAR2,

      no_invalidate    BOOLEAN,

      force            BOOLEAN);

参数说明:

ownname:要分析表的拥有者

tabname:要分析的表名.

partname:分区的名字,只对分区表或分区索引有用.

estimate_percent:采样行的百分比,取值范围[0.000001,100],null为全部分析,不采样. 常量:DBMS_STATS.AUTO_SAMPLE_SIZE是默认值,由oracle决定最佳取采样值.

block_sample:是否用块采样代替行采样.

method_opt:决定histograms信息是怎样被统计的.method_opt的取值如下(默认值为FOR ALL COLUMNS SIZE AUTO):

for all columns:统计所有列的histograms.

for all indexed columns:统计所有indexed列的histograms.

for all hidden columns:统计你看不到列的histograms

for columns <list> SIZE <N> | REPEAT | AUTO | SKEWONLY:统计指定列的histograms.N的取值范围 [1,254]; REPEAT上次统计过的histograms;AUTO由oracle决定N的大小;SKEWONLY multiple end-points with the same value which is what we define by "there is skew in thedata

degree:决定并行度.默认值为null.

granularity:Granularity of statistics to collect ,only pertinent if the table is partitioned.

cascade:是收集索引的信息.默认为FALSE.

stattab:指定要存储统计信息的表,statid如果多个表的统计信息存储在同一个stattab中用于进行区分.statown存储统计信息表的拥有者.以上三个参数若不指定,统计信息会直接更新到数据字典.

no_invalidate: Does not invalidate the dependent cursors if set to TRUE. The procedure invalidatesthe dependent cursors immediately if set to FALSE.

:即使表锁住了也收集统计信息.