通过append hint来插入数据，演示它和普通插入数据的性能比较。

一、实验说明

操作系统：redhat 5.4

数据库:oracle 11g r2

二、在Noarchivelog模式中的比较：

SQL> conn jack/jack
Connected.
SQL> drop table t purge;

Table dropped.

SQL> conn /as sysdba
Connected.
SQL> archive log list;
Database log mode           No Archive Mode              --Noarchivelog模式
Automatic archival           Disabled
Archive destination           USE_DB_RECOVERY_FILE_DEST
Oldest online log sequence     5
Current log sequence           7
SQL> conn jack/jack
Connected.
SQL> create table t as select * from dba_objects;

Table created.

SQL> set autotrace traceonly;
SQL> insert into t select * from t;

72544 rows created.

Execution Plan
----------------------------------------------------------
Plan hash value: 1601196873

---------------------------------------------------------------------------------
| Id  | Operation         | Name | Rows    | Bytes | Cost (%CPU)| Time    |
---------------------------------------------------------------------------------
|   0 | INSERT STATEMENT     |    |   285K|    56M|   590   (1)| 00:00:08 |
|   1 |  LOAD TABLE CONVENTIONAL | T    |    |    |         |        |
|   2 |   TABLE ACCESS FULL     | T    |   285K|    56M|   590   (1)| 00:00:08 |
---------------------------------------------------------------------------------

Note
-----
- dynamic sampling used for this statement (level=2)

Statistics
----------------------------------------------------------
586      recursive calls
9343     db block gets
2449     consistent gets
1033     physical reads
8428664  redo size
673      bytes sent via SQL*Net to client
601      bytes received via SQL*Net from client
3       SQL*Net roundtrips to/from client
2       sorts (memory)
0       sorts (disk)
72544    rows processed

SQL> set autotrace off;
SQL> drop table t purge;

Table dropped.

SQL> create table t as select * from dba_objects;

Table created.

SQL> set autotrace traceonly;
SQL> insert /*+ append */ into t select * from t;

72544 rows created.

Execution Plan
----------------------------------------------------------
ERROR:
ORA-12838: cannot read/modify an object after modifying it in parallel

SP2-0612: Error generating AUTOTRACE EXPLAIN report

Statistics
----------------------------------------------------------
594     recursive calls
1292    db block gets
1218    consistent gets
1033    physical reads
22472   redo size
662     bytes sent via SQL*Net to client
615     bytes received via SQL*Net from client
3      SQL*Net roundtrips to/from client
2      sorts (memory)
0      sorts (disk)
93     72544   rows processed

小结：

2.1、append减少数据块读的数量，普通插入db block gets为9343，加/*+ append */为1292；

2.2、append减少一致读的数量，普通插入consistent gets为2449，加/*+ append */为1218；

2.3、append减少回滚的写入数量，普通插入redo size为8428664，加/*+ append */为22472。

三、在Arachivelog模式下比较(前提是不使用alter table t nologging;)

SQL> conn /as sysdba
Connected.
SQL> shutdown immediate;
Database closed.
Database dismounted.
ORACLE instance shut down.
SQL> startup mount;
ORACLE instance started.

Total System Global Area  372449280 bytes
Fixed Size            1336624 bytes
Variable Size          146803408 bytes
Database Buffers      218103808 bytes
Redo Buffers            6205440 bytes
Database mounted.
SQL> alter database archivelog;

Database altered.

SQL> alter database open;

Database altered.

SQL> archive log list;
Database log mode           Archive Mode
Automatic archival           Enabled
Archive destination           USE_DB_RECOVERY_FILE_DEST
Oldest online log sequence     5
Next log sequence to archive   7
Current log sequence           7
SQL> conn jack/jack
Connected.
SQL> drop table t purge;

Table dropped.

SQL> create table t as select * from dba_objects;

Table created.

SQL> set linesize 160;
SQL> set autotrace traceonly;
SQL> insert into t select * from t;

72544 rows created.

Execution Plan
----------------------------------------------------------
Plan hash value: 1601196873

---------------------------------------------------------------------------------
| Id  | Operation         | Name | Rows    | Bytes | Cost (%CPU)| Time    |
---------------------------------------------------------------------------------
|   0 | INSERT STATEMENT     |    |   285K|    56M|   590   (1)| 00:00:08 |
|   1 |  LOAD TABLE CONVENTIONAL | T    |    |    |         |        |
|   2 |   TABLE ACCESS FULL     | T    |   285K|    56M|   590   (1)| 00:00:08 |
---------------------------------------------------------------------------------

Note
-----
- dynamic sampling used for this statement (level=2)

Statistics
----------------------------------------------------------
586      recursive calls
9341     db block gets
2395     consistent gets
1033     physical reads
8428488  redo size
679      bytes sent via SQL*Net to client
601      bytes received via SQL*Net from client
3       SQL*Net roundtrips to/from client
2       sorts (memory)
0       sorts (disk)
72544    rows processed

SQL> set autotrace off;
SQL> drop table t purge;

Table dropped.

SQL> create table t as select * from dba_objects;

Table created.

SQL> set autotrace traceonly;
SQL> insert /*+ append */ into t select * from t;

72544 rows created.

Execution Plan
----------------------------------------------------------
ERROR:
ORA-12838: cannot read/modify an object after modifying it in parallel

SP2-0612: Error generating AUTOTRACE EXPLAIN report

Statistics
----------------------------------------------------------
594      recursive calls
1292     db block gets
1218     consistent gets
1033     physical reads
8536868  redo size
665      bytes sent via SQL*Net to client
615      bytes received via SQL*Net from client
3       SQL*Net roundtrips to/from client
2       sorts (memory)
0       sorts (disk)
72544     rows processed

小结：

3.1、append减少数据块读的数量，普通插入db block gets为9341，加append为1292；

3.2、append减少一致读的数量，普通插入consistents gets为2395，加append为1218；

3.3、回滚段的数量没有什么变化。

四、在Arachivelog模式下比较(前提是使用alter table t nologging;)

SQL> conn /as sysdba
Connected.
SQL> archive log list;
Database log mode           Archive Mode
Automatic archival           Enabled
Archive destination           USE_DB_RECOVERY_FILE_DEST
Oldest online log sequence     7
Next log sequence to archive   9
Current log sequence           9

SQL> drop table t purge;

Table dropped.

SQL> create table t as select * from dba_objects;

Table created.

SQL> alter table t nologging;

Table altered.

SQL> set autotrace traceonly;
SQL> insert into t select * from t;

72544 rows created.

Execution Plan
----------------------------------------------------------
Plan hash value: 1601196873

---------------------------------------------------------------------------------
| Id  | Operation         | Name | Rows    | Bytes | Cost (%CPU)| Time    |
---------------------------------------------------------------------------------
|   0 | INSERT STATEMENT     |    |   285K|    56M|   590   (1)| 00:00:08 |
|   1 |  LOAD TABLE CONVENTIONAL | T    |    |    |         |        |
|   2 |   TABLE ACCESS FULL     | T    |   285K|    56M|   590   (1)| 00:00:08 |
---------------------------------------------------------------------------------

Note
-----
- dynamic sampling used for this statement (level=2)

Statistics
----------------------------------------------------------
766      recursive calls
9347     db block gets
2352     consistent gets
1033     physical reads
8434036  redo size
681      bytes sent via SQL*Net to client
601      bytes received via SQL*Net from client
3       SQL*Net roundtrips to/from client
6       sorts (memory)
0       sorts (disk)
58     72544    rows processed

SQL> drop table t purge;

Table dropped.

SQL> set autotrace off;
SQL> create table t as select * from dba_objects;

Table created.

SQL> alter table t nologging;

Table altered.

SQL> set autotrace traceonly;
SQL> insert /*+ append */ into t select * from t;

72544 rows created.

Execution Plan
----------------------------------------------------------
ERROR:
ORA-12838: cannot read/modify an object after modifying it in parallel

SP2-0612: Error generating AUTOTRACE EXPLAIN report

Statistics
----------------------------------------------------------
774    recursive calls
1292   db block gets
1237   consistent gets
1033   physical reads
22472  redo size
666    bytes sent via SQL*Net to client
615    bytes received via SQL*Net from client
3     SQL*Net roundtrips to/from client
6     sorts (memory)
0     sorts (disk)
99     72544  rows processed

小结：

4.1、append减少数据块的数量，普通插入db block gets为9347，加append为1292；

4.2、append减少一致读的数量，普通插入consistent gets为2352，加append为1237；

4.3、append减少回滚段的写入数量，普通插入redo size为8434036，加append为22472。

五、总结

5.1、只有在归档模式下，对Logging表插入时，加不加append提示，产生的回滚段数量都差不多，其他情况下，加append能显著减少回滚段的产生。

5.2、不论是在什么情况下，加append都能减少数据块读的数量和一致读的数量。