一、引言：

Oracle 9i引入pga_aggregate_target，可以自动对PGA进行调整；

Oracle 10g引入sga_target，可以自动对SGA进行调整；

Oracle 11g则对这两部分进行综合，引入memory_target，可以自动调整所有的内存，这就是新引入的自动内存管理特性。

二、本文说明：

操作系统：rhel 5.4 x32

数据库：oracle 11g r2

三、memory_target的介绍：

3.1、下面通过示例了解一下memory_target的设置与PGA和SGA关系：

1 SQL> show parameter memory_target;

2

3 NAME TYPE VALUE

4 ------------------------------------ ----------- ------------------------------

5 memory_target big integer 316M

6 SQL> show sga;

7

8 Total System Global Area 330600448 bytes

9 Fixed Size 1336344 bytes

10 Variable Size 247466984 bytes

11 Database Buffers 75497472 bytes

12 Redo Buffers 6299648 bytes

13 SQL> alter system set memory_target=200m scope=spfile;

14

15 System altered.

16

17 SQL> alter system set sga_target=0 scope=spfile;

18

19 System altered.

20

21 SQL> alter system set pga_aggregate_target=0 scope=spfile;

22

23 System altered.

24

25 SQL> shutdown immediate;

26 Database closed.

27 Database dismounted.

28 ORACLE instance shut down.

29 SQL> startup

30 ORACLE instance started.

31

32 Total System Global Area 209235968 bytes

33 Fixed Size 1335528 bytes

34 Variable Size 201330456 bytes

35 Database Buffers 4194304 bytes

36 Redo Buffers 2375680 bytes

37 Database mounted.

38 Database opened.

设置memory_target参数后，实际上Oracle会自动设置并调整一下两个参数来分配SGA和PGA的内存，这和Oracle 10g自动设置sga_target后分配db_cache_size和shared_pool_size的机制是一样的。

1 SQL> col ksppinm for a20;

2 SQL> col ksppstvl for a20;

3 SQL> select a.ksppinm name,b.ksppstvl value

4 2 from x$ksppi a,x$ksppcv b

5 3 where a.indx = b.indx

6 4 and (a.ksppinm like '%sga_target%'

7 5 or a.ksppinm like '%pga_aggregate_target%');

8

9 NAME VALUE

10 ----------------------- -------------------

11 sga_target 0

12 __sga_target 142606336

13 pga_aggregate_target 0

14 __pga_aggregate_target 67108864

3.2、讨论一下11g中memory_target设置和不设置对SGA/PGA的影响：

3.2.1、如果memory_target设置为非0值

(下面有四种情况来对SGA和PGA的大小进行分配)

3.2.1.1、sga_target和pga_aggregate_target已经设置大小

如果Oracle中已经设置了参数sga_target和pga_aggregate_target，则这两个参数将各自被分配为最小值为他们的目标值。

memory_target = sga_target + pga_aggregate_target，大小和memory_max_size一致。

3.2.1.2、sga_target设置大小，pga_aggregate_target没有设置大小

那么pga_aggregate_target初始化值=memory_target-sga_target

3.2.1.3、sga_target没有设置大小，pga_aggregate_target设置大小

那么sga_target初始化值=memory_target-pga_aggregate_target

3.2.1.4、sga_target和pga_aggregate_target都没有设置大小

Oracle 11g中对这种sga_target和pag_aggregate_target都没有设定大小的情况下，Oracle将对这两个值没有最小值和默认值。Oracle将根据数据库运行状况进行分配大小。但在数据库启动是会有一个固定比例来分配：

sga_target = memory_target*60%

pga_aggregate_target = memory_target*40%

3.2.2、如果memory_target没有设置或 = 0(在11g中默认为0)

11g中默认为0则初始状态下取消了memory_target的作用，完全和10g在内存管理上一致，完全向下兼容。(也有三种情况来对SGA和PGA的大小进行分配)

3.2.2.1、sga_target设置值，则自动调节SGA中的shared pool,buffer cache,redo log buffer,java pool,larger pool等内存空间的大小。PGA则依赖pga_aggregate_target的大小。sga和pga不能自动增长和自动缩小。

3.2.2.2、sga_target和pga_aggregate_target都没有设置

SGA中的各组件大小都要明确设定，不能自动调整各组件大小。PGA不能自动增长和收缩。

3.2.2.3、memory_max_target设置而memory_target = 0这种情况先和10g一样，不做说明。

三、ORA-00845

最后谈一下ORA-00845的由来和解决方案：

如果memory_max_target/memory_target设置过大，可能导致instance无法启动，报ORA-00845错误。

1 [oracle@yft bin]$ oerr ora 00845

2 00845, 00000, "MEMORY_TARGET not supported on this system"

3 // *Cause: The MEMORY_TARGET parameter was not supported on this operating system or /dev/shm was not sized

4 // *Action: Refer to documentation for a list of supported operating systems. Or, size /dev/shm to be at leacle instance running on the system.

5

6 SQL> alter system set memory_max_target=400m scope=spfile;

7

8 System altered.

9

10 SQL> shutdown immediate;

11 Database closed.

12 Database dismounted.

13 ORACLE instance shut down.

14 SQL> startup

15 ORA-00845: MEMORY_TARGET not supported on this system

这个错误有点误导，实际上这并不是说该平台版本上不支持AMM特性，只是设置的memory_max_target超过了系统中设置的share memory(/dev/shm)而已。

[oracle@yft bin]$ df -h

Filesystem Size Used Avail Use% Mounted on

tmpfs 395M 0 395M 0% /dev/shm

在Oracle 11g for linux中似乎是用了一种新的机制来管理共享内存段，而不是传统的sys /dev/shm了。在alert.ora中可以找到更准确的错误描述：

1 Mon Feb 25 12:13:21 2013

2 Starting ORACLE instance (normal)

3 WARNING: You are trying to use the MEMORY_TARGET feature. This feature requires the /dev/shm file system to be mounted for at least 419430400 bytes. /dev/shm is either not mounted or is mounted with available space less than this size. Please fix this so that MEMORY_TARGET can work as expected. Current available is 413466624 and used is 0 bytes. Ensure that the mount point is /dev/shm for this directory.

4 memory_target needs larger /dev/shm

解决的办法之一是增加tmpfs文件系统的容量:

修改/etc/fstab中tmpfs对应的行；将原来的tmpfs   /dev/shm  tmpfs defaults 0 0 改成tmpfs /dev/shm tmpfs default,size=1024M 0 0,这样tmpfs增大为1G，重新mount /dev/shm使之生效。

1 [root@yft ~]# vi /etc/fstab

2 tmpfs /dev/shm tmpfs defaults,size=420m 0 0

3

4 [root@yft ~]# mount -o remount /dev/shm

5 [root@yft ~]# df -h

6 Filesystem Size Used Avail Use% Mounted on

7 tmpfs 420M 0 420M 0% /dev/shm

8

9 SQL> startup

10 ORACLE instance started.

11

12 Total System Global Area 418484224 bytes

13 Fixed Size 1336932 bytes

14 Variable Size 406849948 bytes

15 Database Buffers 4194304 bytes

16 Redo Buffers 6103040 bytes

17 Database mounted.

18 Database opened.