使用dbms_stats.gather_table_stats调整表的统计信息

创建实验表，插入10万行数据

SQL> create table test (id number,name varchar2(10));

Table created.

SQL> declare
begin
for i in 1..100000 loop
insert into test values(1,'a');
commit;
end loop;
end;
/ 

PL/SQL procedure successfully completed.

SQL> commit;

Commit complete.

查看表的统计信息，统计信息为空

SQL> select table_name ,num_rows,blocks,avg_row_len from user_tables where table_name='TEST';

TABLE_NAME NUM_ROWS BLOCKS AVG_ROW_LEN

------------------------------ ---------- ---------- -----------
TEST

收集表的统计信息

SQL> exec dbms_stats.gather_table_stats('SCOTT','TEST');

PL/SQL procedure successfully completed.

SQL> select table_name,num_rows,blocks,avg_row_len from user_tables where table_name='TEST';

TABLE_NAME 　　　　 NUM_ROWS BLOCKS AVG_ROW_LEN
------------------------------ ---------- ---------- -----------
TEST 　　　　　　　　 100000　　 244 　　 5

查看全表扫描的执行计划与cost

SQL> set lines 300 pages 300
SQL> explain plan for select * from test;

Explained.

SQL> select * from table(dbms_xplan.display());

PLAN_TABLE_OUTPUT
----------------------------------------------------------------------------------
Plan hash value: 1357081020

--------------------------------------------------------------------------
| Id | Operation | Name | Rows | Bytes | Cost (%CPU)| Time |
--------------------------------------------------------------------------
| 0 | SELECT STATEMENT | | 100K| 488K| 69 (2)| 00:00:01 |
| 1 | TABLE ACCESS FULL| TEST | 100K| 488K| 69 (2)| 00:00:01 |
--------------------------------------------------------------------------

8 rows selected.

计算将行数由10万提高到200万所使用的块数

SQL> select round(244*2000000/100000) from dual;

ROUND(244*2000000/100000)
-------------------------
4880

通过使用set_table_stats修改统计信息

SQL> exec dbms_stats.set_table_stats(ownname=>'SCOTT',tabname=>'TEST',numrows=>2000000,numblks=>4880,avgrlen=>5);

PL/SQL procedure successfully completed.

SQL>select table_name,num_rows,blocks,avg_row_len from user_tables where table_name='TEST';

TABLE_NAME NUM_ROWS BLOCKS AVG_ROW_LEN
------------------------------ ---------- ---------- -----------
TEST 　　　　2000000 　　　　4880 　　 5

查看全表扫描的执行计划与cost

SQL> explain plan for select * from test;

Explained.

SQL> select * from table(dbms_xplan.display());

PLAN_TABLE_OUTPUT
-------------------------------------------------------------------------------------
Plan hash value: 1357081020

--------------------------------------------------------------------------
| Id | Operation | Name | Rows | Bytes | Cost (%CPU)| Time |
--------------------------------------------------------------------------
| 0 | SELECT STATEMENT | | 2000K| 9765K| 1341 (2)| 00:00:17 |
| 1 | TABLE ACCESS FULL| TEST | 2000K| 9765K| 1341 (2)| 00:00:17 |
--------------------------------------------------------------------------

8 rows selected.