alter table move的用途

降低HWM，消除行移植和行链接。

我们可以通过user_segments 或 user_extents 的blocks来查看hwm

SQL> select segment_name,blocks
  2  from user_segments
  3  where segment_name='SMTDZ';

SEGMENT_NAME         BLOCKS
---------------- ----------
SMTDZ                    16

在usr_extents中查找

SQL> select segment_name,blocks
  2  from user_extents where segment_name='SMTDZ';

SEGMENT_NAME         BLOCKS
---------------- ----------
SMTDZ                     8
SMTDZ                     8
SQL> select count(*) from smtdz;

  COUNT(*)
----------
      1015

SQL> insert into smtdz select * from smtdz;

已创建1015行。

SQL> commit;

提交完成。

SQL> select count(*) from smtdz;

  COUNT(*)
----------
      2030

SQL> select segment_name,blocks
  2  from user_segmentS
  3  where segment_name='SMTDZ';

SEGMENT_NAME         BLOCKS
---------------- ----------
SMTDZ                    24

SQL> select segment_name,blocks from user_extents where segment_name='SMTDZ';

SEGMENT_NAME         BLOCKS
---------------- ----------
SMTDZ                     8
SMTDZ                     8
SMTDZ                     8

可以看到当我们insert into smtdz 一批数据后其blocks增加了，我们再对smtdz进行delete下

SQL> delete from smtdz where rownum<1001;

已删除1000行。

SQL> commit;

提交完成。

SQL> select count(*) from smtdz;

  COUNT(*)
----------
      1030

未对表smtdz进行分析analyze前

SQL> select segment_name,blocks
  2  from user_segments
  3  where segment_name='SMTDZ';

SEGMENT_NAME         BLOCKS
---------------- ----------
SMTDZ                    24

对表smtdz进行分析

SQL> exec dbms_stats.gather_table_stats(ownname=>'IC',tabname=>'SMTDZ');

PL/SQL 过程已成功完成。

SQL> select segment_name,blocks
  2  from user_segments
  3  where segment_name='SMTDZ';

SEGMENT_NAME         BLOCKS
---------------- ----------
SMTDZ                    24

可见分析表只是修改了user_tables里的num_rows值，对hwm并没有影响

SQL> alter table smtdz move;

表已更改。

SQL> select segment_name,blocks
  2  from user_segments
  3  where segment_name='SMTDZ';

SEGMENT_NAME         BLOCKS
---------------- ----------
SMTDZ                    16

从这里可以看到表smtdz的hwm发生了变化，现在只有16个blocks了

 

我们可以用下面的办法来最方便的进行重组，消除row migration：

SQL> alter table t add t1 date default sysdate;
Table altered.

SQL> c/t1/t2
1* alter table t add t2 date default sysdate
SQL> /
Table altered.

SQL> c/t2/t3
1* alter table t add t3 date default sysdate
SQL> /
Table altered.

SQL> analyze table t compute statistics;
Table analyzed.

SQL> select table_name,num_rows,CHAIN_CNT from user_tables where table_name='T';
TABLE_NAME  NUM_ROWS    CHAIN_CNT
-------------  --------------- ---------- 
T                   41616              3908

SQL> alter table t move ;
Table altered.

SQL> analyze table t compute statistics;
Table analyzed.

SQL> select table_name,num_rows,CHAIN_CNT from user_tables where table_name='T';
TABLE_NAME NUM_ROWS      CHAIN_CNT
-------------  ---------------- ------------
T                   41616                0