Oracle 收集统计值 收集统计信息

--收集Oracle统计信息
BEGIN
  DBMS_STATS.GATHER_TABLE_STATS(OWNNAME          => 'OWNER_NAME',  --数据库用户名
                                TABNAME          => 'TABLE_NAME',  --表名
                                PARTNAME         => 'PARTITION_NAME',  --分区名
                                ESTIMATE_PERCENT => DBMS_STATS.AUTO_SAMPLE_SIZE,
                                METHOD_OPT       => 'for all columns size repeat',
                                DEGREE           => DBMS_STATS.AUTO_DEGREE,
                                CASCADE          => TRUE);
END;

--查询表的统计信息
SELECT *
  FROM DBA_TAB_STATISTICS S
 WHERE S.OWNER = 'OWNER_NAME'   --数据库用户名
   AND S.TABLE_NAME = 'TABLE_NAME'  --表名
 ORDER BY S.LAST_ANALYZED DESC;


--Oracle统计信息
DBMS_STATS.gather_table_stats
    (ownname varchar2,
     tabname varchar2,
     partname varchar2 default null,
     estimate_percent number default   to_estimate_percent_type(get_param('ESTIMATE_PERCENT')),
     block_sample boolean default FALSE,
     method_opt varchar2 default get_param('METHOD_OPT'),
     degree number default to_degree_type(get_param('DEGREE')),
     granularity varchar2 default get_param('GRANULARITY'),
     cascade boolean default to_cascade_type(get_param('CASCADE')),
     stattab varchar2 default null, statid varchar2 default null,
     statown varchar2 default null,
     no_invalidate boolean default to_no_invalidate_type(get_param('NO_INVALIDATE')),
     stattype varchar2 default 'DATA',
     force boolean default FALSE);

/*
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_sapmple: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:
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]; R
                                              EPEAT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 the data
degree:              设置收集统计信息的并行度.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falase.
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 invalidates the dependent cursors 
                        immediately if set to FALSE.
force:         即使表锁住了也收集统计信息

*/
--例子:
execute dbms_stats.gather_table_stats(ownname => 'owner',
                                      tabname => 'table_name' ,
                                      estimate_percent => null ,
                                      method_opt => 'for all indexed columns' ,
                                      cascade => true);
--GATHER_INDEX_STATS
--==========================
BEGIN
SYS.DBMS_STATS.GATHER_INDEX_STATS (OwnName => 'ABC',
                                   IndName => 'IDX_FUNC_ABC',
                                   Estimate_Percent => 10,
                                   Degree => SYS.DBMS_STATS.DEFAULT_DEGREE,
                                   No_Invalidate => FALSE);
END;