ORACLE 10g append跟nologging插入对REDO的影响

/*+append*/ 

1.  append 属于direct insert,归档模式下append+table nologging会大量减少日志,

    非归档模式append会大量减少日志,append方式插入只会产生很少的undo

2.

 

----------------------------------------------------------------------------------------------------------------------------------------------------

 

oracle append有什么作用？

 

----------------------------------------------------------------------------------------------------------------------------------------------------

 

 

 

SQL> select * from v$version;

BANNER

----------------------------------------------------------------

Oracle Database 10g Enterprise Edition Release 10.2.0.1.0 - Prod

PL/SQL Release 10.2.0.1.0 - Production

CORE    10.2.0.1.0      Production

TNS for 32-bit Windows: Version 10.2.0.1.0 - Production

NLSRTL Version 10.2.0.1.0 – Production

1.非归档模式append,nologging,append+nologging 三种情况数据产生REDO的对比

PHP:

SQL> archive log list
Database log mode              No Archive Mode
Automatic archival             Disabled
Archive destination            USE_DB_RECOVERY_FILE_DEST
Oldest online log sequence     13
Current log sequence           15

SQL> create table t as select * from dba_objects where 1=2;

Table created.

SQL> select name,value,class from v$sysstat where name='redo size';

NAME            VALUE      CLASS
---------- ---------- ----------
redo size      582728          2

SQL> insert into t select * from dba_objects;

10947 rows created.

SQL> select name,value,class from v$sysstat where name='redo size';

NAME            VALUE      CLASS
---------- ---------- ----------
redo size     1745704          2

SQL> insert /*+append*/ into t select * from dba_objects;

10947 rows created.

SQL> select name,value,class from v$sysstat where name='redo size';

NAME            VALUE      CLASS
---------- ---------- ----------
redo size     1839872          2

SQL> select (1745704-582728) redo1,(1839872-1745704) redo2 from dual;

     REDO1      REDO2
---------- ----------
   1162976     94168

SQL> drop table t;

Table dropped.

SQL> create table t nologging as select * from dba_objects where 1=2;

Table created.

SQL> select name,value,class from v$sysstat where name='redo size';

NAME            VALUE      CLASS
---------- ---------- ----------
redo size     3441836          2

SQL> insert into t select * from dba_objects;

10947 rows created.

SQL> select name,value,class from v$sysstat where name='redo size';

NAME            VALUE      CLASS
---------- ---------- ----------
redo size     4660204          2

SQL> insert /*+append*/ into t select * from dba_objects;

10947 rows created.

SQL> select name,value,class from v$sysstat where name='redo size';

NAME            VALUE      CLASS
---------- ---------- ----------
redo size     4667180          2

SQL> select (4660204-3441836) redo1,(4667180-4660204) redo2 from dual;

     REDO1      REDO2
---------- ----------
   1218368       6976

2.归档下:append,nologging,append+nologging 三种情况数据产生REDO的对比

PHP语言:SQL> select flashback_on from v$database;

FLASHBACK_ON
------------------
NO

SQL> archive log list
Database log mode              Archive Mode
Automatic archival             Enabled
Archive destination            USE_DB_RECOVERY_FILE_DEST
Oldest online log sequence     13
Next log sequence to archive   15
Current log sequence           15
SQL> select log_mode from v$database;

LOG_MODE
------------
ARCHIVELOG

SQL> create table t as select * from dba_objects where 1=2;

Table created.

SQL> select name,value from v$sysstat where name='redo size';

NAME            VALUE
---------- ----------
redo size      155624

SQL> insert into t select * from dba_objects;

10947 rows created.

SQL> select name,value from v$sysstat where name='redo size';

NAME            VALUE
---------- ----------
redo size     1316420

SQL> insert /*+append*/ into t select * from dba_objects;

10947 rows created.

SQL> select name,value from v$sysstat where name='redo size';

NAME            VALUE
---------- ----------
redo size     2461876

SQL> select (1316420-155624) redo1,(2461876-1316420) redo2 from dual;

     REDO1      REDO2
---------- ----------
   1160796    1145456

SQL> truncate table t;

Table truncated.

SQL> alter table t nologging;

Table altered.

SQL> select name,value from v$sysstat where name='redo size';

NAME            VALUE
---------- ----------
redo size     2505320

SQL> insert into t select * from dba_objects;

10947 rows created.

SQL> select name,value from v$sysstat where name='redo size';

NAME            VALUE
---------- ----------
redo size     3667856

SQL> insert /*+append*/ into t select * from dba_objects;

10947 rows created.

SQL> select name,value from v$sysstat where name='redo size';

NAME            VALUE
---------- ----------
redo size     3670424

SQL> select (3667856-2505320) redo1,(3670424-3667856) redo2 from dual;

     REDO1      REDO2
---------- ----------
   1162536       2568

总结:

 

 normal
 append
 nologging
 Append+nologging
 
 
Noarchive
 1162976
 94168
 1218368
 6976
 
 
Archive
 1160796
 1145456
 1162536
 2568
 
 

 

可以看出

1. 不管哪种模式下append要与nologging方式联用才能达到很好的效果。

2. 非归档与归档方式，只用NOLOGGING是不起效果的。

3. 非归档下append已达到不错的效果，但不及与nologging的联用方式。

4. 归档下单append起不到效果。

NOLOGGING插完后最好做个备份。

另外，如果库处在FORCELOGGING模式下，此时的nologging方式是无效的，这个我也测试过。

ITPUB上也有关于NOLOGGING何时生效的讨论

http://www.itpub.net/showthread.php?threadid=239905

eygle也做过这个实验

http://www.eygle.com/faq/Nologging&append.htm

 

 

 

Originally posted by jwzl at 2004-12-6 09:02:
我想确认一下/*+append*/系统到底会不会写日志，
很多人说不写日志，但怎么还有另外一个参数nologing,
如果真的没有写日志，也是很麻烦的

写的日志量不同^_^.
汇总一下下面的例子中关于redo log的信息.
pure insert                587624  
insert with nologging  585496  
insert with append 2240  
insert with append & nologging 400  

[quote]
Connected to:
Oracle9i Enterprise Edition Release 9.2.0.4.0 - Production
With the Partitioning, OLAP and Oracle Data Mining options
JServer Release 9.2.0.4.0 - Production

SQL> drop table t;

Table dropped.
SQL> create table t nologging as select * from dba_objects where null = null;

Table created.

SQL> set autot traceonly stat
SQL> insert into t select * from dba_objects;

5888 rows created.

Statistics
----------------------------------------------------------
        271  recursive calls
        773  db block gets
      12653  consistent gets
          0  physical reads
     587624  redo size
        617  bytes sent via SQL*Net to client
        539  bytes received via SQL*Net from client
          3  SQL*Net roundtrips to/from client
          2  sorts (memory)
          0  sorts (disk)
       5888  rows processed

SQL> insert into t nologging select * from dba_objects;

5888 rows created.

Statistics
----------------------------------------------------------
         28  recursive calls
        721  db block gets
      12654  consistent gets
          0  physical reads
     585496  redo size
        618  bytes sent via SQL*Net to client
        549  bytes received via SQL*Net from client
          3  SQL*Net roundtrips to/from client
          1  sorts (memory)
          0  sorts (disk)
       5888  rows processed

SQL> insert /*+append*/ into t select * from dba_objects;

5888 rows created.

Statistics
----------------------------------------------------------
         29  recursive calls
         31  db block gets
      12526  consistent gets
          0  physical reads
       2240  redo size
        603  bytes sent via SQL*Net to client
        551  bytes received via SQL*Net from client
          3  SQL*Net roundtrips to/from client
          1  sorts (memory)
          0  sorts (disk)
       5888  rows processed

SQL> commit;

Commit complete.

SQL> insert /*+append*/ into t nologging select * from dba_objects;

5888 rows created.

Statistics
----------------------------------------------------------
          7  recursive calls
          8  db block gets
      12517  consistent gets
          0  physical reads
        400  redo size
        603  bytes sent via SQL*Net to client
        561  bytes received via SQL*Net from client
          3  SQL*Net roundtrips to/from client
          1  sorts (memory)
          0  sorts (disk)
       5888  rows processed

SQL> commit;

Commit complete.

SQL> exit