海量数据备份和恢复方案
对于OLAP的数据库的业务特点，是将批量的数据加载入库，然后对这些数据进行分析处理，比如报表或者数据挖掘，最后给业务提供一种决策支持；另外，这类数据库的数据实时性非常高，一旦这些数据处理完毕后，就很少再次使用（有时，也需要对这类数据进行查询）。
对于OLAP数据库的备份和恢复可以考虑这样几种方案：
1、使用分布式数据库
将数据分布到多个库里，当数据库恢复时，只需要恢复单个库的数据，大大节省恢复时间。
2、结合分区技术，以传输表空间方式进行备份和恢复
1、建立分区表，将分区存储在不同的表空间
[[email protected] ~]$sqlplus '/as sysdba'
SQL*Plus: Release 11.2.0.1.0 Production on Tue Nov 18 17:15:47 2014
Copyright (c) 1982, 2009, Oracle. All rights reserved.
Connected to:
Oracle Database 11g Enterprise Edition Release 11.2.0.1.0 - Production
With the Partitioning, OLAP, Data Mining and Real Application Testing options
17:15:47 [email protected] prod >create tablespace tbs1
17:16:03 2 datafile '/dsk1/oradata/prod/tbs1.dbf' size 10m;
Tablespace created.
17:17:00 [email protected] prod >create tablespace tbs2
17:17:11 2 datafile '/dsk2/oradata/prod/tbs2.dbf' size 10m;
Tablespace created.
17:17:49 [email protected] prod >create tablespace tbs3
17:17:57 2 datafile '/dsk3/oradata/prod/tbs3.dbf' size 10m;
Tablespace created.
17:18:35 [email protected] prod >create tablespace tbs1_indx
17:18:49 2 datafile '/dsk1/oradata/prod/tbs1_indx.dbf' size 10m;
Tablespace created.
17:19:43 [email protected] prod >create tablespace tbs2_indx
17:19:54 2 datafile '/dsk2/oradata/prod/tbs2_indx.dbf' size 10m;
Tablespace created.
17:20:18 [email protected] prod >create tablespace tbs3_indx
17:20:30 2 datafile '/dsk3/oradata/prod/tbs3_indx.dbf' size 10m;
Tablespace created.
17:22:12 [email protected] prod >select file_id,file_name,tablespace_name from dba_data_files
FILE_ID FILE_NAME TABLESPACE_NAME
---------- -------------------------------------------------- ------------------------------
11 /dsk1/oradata/prod/tbs1.dbf TBS1
12 /dsk2/oradata/prod/tbs2.dbf TBS2
13 /dsk3/oradata/prod/tbs3.dbf TBS3
4 /u01/app/oracle/oradata/prod/users01.dbf USERS
3 /u01/app/oracle/oradata/prod/undotbs01.dbf UNDOTBS1
2 /u01/app/oracle/oradata/prod/sysaux01.dbf SYSAUX
1 /u01/app/oracle/oradata/prod/system01.dbf SYSTEM
5 /u01/app/oracle/oradata/prod/example01.dbf EXAMPLE
6 /u01/app/oracle/oradata/prod/users02.dbf USERS
7 /u01/app/oracle/oradata/prod/catatbs1.dbf CATATBS
8 /u01/app/oracle/oradata/prod/perfertbs1.dbf PERFERTBS
9 /u01/app/oracle/oradata/prod/oggtbs1.dbf OGG_TBS
10 /u01/app/oracle/oradata/prod/test1.dbf TEST1
14 /dsk1/oradata/prod/tbs1_indx.dbf TBS1_INDX
15 /dsk2/oradata/prod/tbs2_indx.dbf TBS2_INDX
16 /dsk3/oradata/prod/tbs3_indx.dbf TBS3_INDX
建立分区表及索引：
17:26:41 [email protected] prod >create table t1(id int,name varchar2(1000))
17:26:57 2 partition by range(id)
17:27:01 3 (partition p1 values less than(1000) tablespace tbs1,
17:27:13 4 partition p2 values less than(2000) tablespace tbs2,
17:27:23 5 partition p3 values less than(maxvalue) tablespace tbs3);
Table created.
17:30:33 [email protected] prod >create index t1_indx on t1(id) local
2 (
3 partition p1 tablespace tbs1_indx,
4 partition p2 tablespace tbs2_indx,
5* partition p3 tablespace tbs3_indx )
/
17:30:37 [email protected] prod >select partition_name,tablespace_name from user_segments where segment_name='T1';
PARTITION_NAME TABLESPACE_NAME
------------------------------ ------------------------------
P1 TBS1
P2 TBS2
P3 TBS3
17:31:33 [email protected] prod >select partition_name,tablespace_name from user_segments where segment_name='T1_INDX';
PARTITION_NAME TABLESPACE_NAME
------------------------------ ------------------------------
P1 TBS1_INDX
P2 TBS2_INDX
P3 TBS3_INDX
插入数据：
17:34:09 [email protected] prod >begin
17:34:26 2 for i in 1..3 loop
17:34:32 3 insert into scott.t1 select object_id*i,object_name from dba_objects where object_id <1000;
17:34:43 4 end loop;
17:34:51 5 commit;
17:34:57 6 end;
17:35:02 7 /
PL/SQL procedure successfully completed.
17:32:08 [email protected] prod >select count(*) from t1;
COUNT(*)
----------
2826
17:36:52 [email protected] prod >select 'p1',count(*) from t1 partition(p1)
17:37:42 2 union
17:37:47 3 select 'p2',count(*) from t1 partition(p2)
17:38:11 4 union
17:38:13 5 select 'p3',count(*) from t1 partition(p3);
'P1' COUNT(*)
-------------------------------- ----------
p1 1740
p2 774
p3 312
2、传输表空间
17:35:04 [email protected] prod >alter tablespace tbs1 read only;
Tablespace altered.
17:41:02 [email protected] prod >alter tablespace tbs1_indx read only;
Tablespace altered.
17:39:14 [email protected] prod >create directory tbs_dir as '/home/oracle/data';
Directory created.
17:40:30 [email protected] prod >grant read,write on directory tbs_dir to scott;
Grant succeeded.
[[email protected] data]$ expdp system/oracle directory=tbs_dir dumpfile=p1.dmp transport_tablespaces=tbs1,tbs1_indx logfile=p1.log
Export: Release 11.2.0.1.0 - Production on Tue Nov 18 17:44:25 2014
Copyright (c) 1982, 2009, Oracle and/or its affiliates. All rights reserved.
Connected to: Oracle Database 11g Enterprise Edition Release 11.2.0.1.0 - Production
With the Partitioning, OLAP, Data Mining and Real Application Testing options
Starting "SYSTEM"."SYS_EXPORT_TRANSPORTABLE_01": system/******** directory=tbs_dir dumpfile=p1.dmp transport_tablespaces=tbs1,tbs1_indx logfile=p1.log
ORA-39123: Data Pump transportable tablespace job aborted
ORA-39187: The transportable set is not self-contained, violation list is
ORA-39901: Partitioned table SCOTT.T1 is partially contained in the transportable set.
ORA-39921: Default Partition (Table) Tablespace USERS for T1 not contained in transportable set.
Job "SYSTEM"."SYS_EXPORT_TRANSPORTABLE_01" stopped due to fatal error at 17:44:49
传输表空间出错，表空间处于非自包含模式：
18:14:47 [email protected] prod >exec dbms_tts.transport_set_check('TBS1',true);
PL/SQL procedure successfully completed.
18:17:49 [email protected] prod >select * from transport_set_violations;
VIOLATIONS
------------------------------------------------------------------------------------------------------------------------
ORA-39921: Default Partition (Table) Tablespace USERS for T1 not contained in transportable set.
.
ORA-39901: Partitioned table SCOTT.T1 is partially contained in the transportable set.
解决方法，需要创建一个临时表和一个临时表索引，将分区和分区索引交换到临时表和临时表索引表空间上，然后到处临时表和临时表索引。由于临时表不是分区表，它们呢所在的表空间符合自包含条件。
17:45:37 [email protected] prod >create table t1_tmp as select * from t1 where 1=3;
Table created.
Elapsed: 00:00:00.20
17:45:58 [email protected] prod >create index t1_tmp_indx on t1_tmp(id);
Index created.
17:46:33 [email protected] prod >select segment_name,tablespace_name from user_segments
17:47:18 2 where segment_name in ('T1_TMP','T1_TMP_INDX');
SEGMENT_NAME TABLESPACE_NAME
--------------------------------------------------------------------------------- ------------------------------
T1_TMP USERS
T1_TMP_INDX USERS
将分区表交换到临时表：
17:48:32 [email protected] prod >alter table t1 exchange partition p1 with table t1_tmp including indexes;
Table altered.
17:49:02 [email protected] prod >select segment_name,tablespace_name from user_segments
17:49:35 2 where segment_name in ('T1_TMP','T1_TMP_INDX');
SEGMENT_NAME TABLESPACE_NAME
--------------------------------------------------------------------------------- ------------------------------
T1_TMP TBS1
T1_TMP_INDX TBS1_INDX
17:50:44 [email protected] prod >exec dbms_tts.transport_set_check('TBS1',true);
PL/SQL procedure successfully completed.
17:51:59 [email protected] prod >select * from transport_set_violations;
no rows selected
已经符合自包含条件
[[email protected] data]$ expdp system/oracle directory=tbs_dir dumpfile=p1.dmp transport_tablespaces=tbs1,tbs1_indx logfile=p1.log
Export: Release 11.2.0.1.0 - Production on Tue Nov 18 17:52:55 2014
Copyright (c) 1982, 2009, Oracle and/or its affiliates. All rights reserved.
Connected to: Oracle Database 11g Enterprise Edition Release 11.2.0.1.0 - Production
With the Partitioning, OLAP, Data Mining and Real Application Testing options
Starting "SYSTEM"."SYS_EXPORT_TRANSPORTABLE_01": system/******** directory=tbs_dir dumpfile=p1.dmp transport_tablespaces=tbs1,tbs1_indx logfile=p1.log
Processing object type TRANSPORTABLE_EXPORT/PLUGTS_BLK
Processing object type TRANSPORTABLE_EXPORT/TABLE
Processing object type TRANSPORTABLE_EXPORT/INDEX
Processing object type TRANSPORTABLE_EXPORT/INDEX_STATISTICS
Processing object type TRANSPORTABLE_EXPORT/POST_INSTANCE/PLUGTS_BLK
Master table "SYSTEM"."SYS_EXPORT_TRANSPORTABLE_01" successfully loaded/unloaded
******************************************************************************
Dump file set for SYSTEM.SYS_EXPORT_TRANSPORTABLE_01 is:
/home/oracle/data/p1.dmp
******************************************************************************
Datafiles required for transportable tablespace TBS1:
/dsk1/oradata/prod/tbs1.dbf
Datafiles required for transportable tablespace TBS1_INDX:
/dsk1/oradata/prod/tbs1_indx.dbf
Job "SYSTEM"."SYS_EXPORT_TRANSPORTABLE_01" successfully completed at 17:54:17
表空间导出成功！
17:56:16 [email protected] prod >select file_name,tablespace_name from dba_data_files where tablespace_name in ('TBS1','TBS1_INDX');
FILE_NAME TABLESPACE_NAME
-------------------------------------------------- ------------------------------
/dsk1/oradata/prod/tbs1.dbf TBS1
/dsk1/oradata/prod/tbs1_indx.dbf TBS1_INDX
[[email protected] ~]$ cp /dsk1/oradata/prod/tbs1* /home/oracle/data
[[email protected] ~]$ ls -lh /home/oracle/data
total 21M
-rw-r----- 1 oracle oinstall 92K Nov 18 17:54 p1.dmp
-rw-r--r-- 1 oracle oinstall 1.4K Nov 18 17:54 p1.log
-rw-r----- 1 oracle oinstall 11M Nov 18 17:57 tbs1.dbf
-rw-r----- 1 oracle oinstall 11M Nov 18 17:57 tbs1_indx.dbf
然后再将表空间的数据文件进行备份，由于表空间传输，只是导出了metadata，所以数据量非常小，速度非常快。
3、数据恢复
17:58:29 [email protected] prod >drop tablespace tbs1 including contents and datafiles;
Tablespace dropped.
17:58:55 [email protected] prod >drop tablespace tbs1_indx including contents and datafiles;
Tablespace dropped.
17:59:12 [email protected] prod >col segment_name for a20
17:59:42 [email protected] prod >col partition_name for a10
17:59:49 [email protected] prod >col tablespace_name for a10
17:59:59 [email protected] prod >select segment_name,partition_name,tablespace_name from dba_segments
18:00:32 2 where segment_name in ('T1','T1_INDX') order by 2;
SEGMENT_NAME PARTITION_ TABLESPACE
-------------------- ---------- ----------
T1 P1 USERS
T1_INDX P1 USERS
T1_INDX P2 TBS2_INDX
T1 P2 TBS2
T1_INDX P3 TBS3_INDX
T1 P3 TBS3
6 rows selected.
拷贝备份数据文件到数据库下，进行数据导入
[[email protected] oradata]$ cp /home/oracle/data/tbs1*.dbf /u01/app/oracle/oradata/prod/
[[email protected] data]$ impdp system/oracle directory=tbs_dir dumpfile=p1.dmp transport_datafiles='/u01/app/oracle/oradata/prod/tbs1.dbf','/u01/app/oracle/oradata/prod/tbs1_indx.dbf' logfile=imp.log
Import: Release 11.2.0.1.0 - Production on Tue Nov 18 18:06:22 2014
Copyright (c) 1982, 2009, Oracle and/or its affiliates. All rights reserved.
Connected to: Oracle Database 11g Enterprise Edition Release 11.2.0.1.0 - Production
With the Partitioning, OLAP, Data Mining and Real Application Testing options
Master table "SYSTEM"."SYS_IMPORT_TRANSPORTABLE_01" successfully loaded/unloaded
Starting "SYSTEM"."SYS_IMPORT_TRANSPORTABLE_01": system/******** directory=tbs_dir dumpfile=p1.dmp transport_datafiles=/u01/app/oracle/oradata/prod/tbs1.dbf,/u01/app/oracle/oradata/prod/tbs1_indx.dbf logfile=imp.log
Processing object type TRANSPORTABLE_EXPORT/PLUGTS_BLK
Processing object type TRANSPORTABLE_EXPORT/TABLE
Processing object type TRANSPORTABLE_EXPORT/INDEX
Processing object type TRANSPORTABLE_EXPORT/INDEX_STATISTICS
Processing object type TRANSPORTABLE_EXPORT/POST_INSTANCE/PLUGTS_BLK
Job "SYSTEM"."SYS_IMPORT_TRANSPORTABLE_01" successfully completed at 18:06:37
数据导入成功
18:01:03 [email protected] prod >select segment_name,partition_name,tablespace_name from dba_segments
18:07:37 2 where segment_name in ('T1_TMP','T1_TMP_INDX');
SEGMENT_NAME PARTITION_ TABLESPACE
-------------------- ---------- ----------
T1_TMP TBS1
T1_TMP_INDX TBS1_INDX
18:09:40 [email protected] prod >alter table t1 exchange partition p1 with table t1_tmp including indexes;
Table altered.
18:08:15 [email protected] prod >select segment_name,partition_name,tablespace_name from dba_segments
18:10:46 2 where segment_name in ('T1','T1_INDX') order by 2;
SEGMENT_NAME PARTITION_ TABLESPACE
-------------------- ---------- ----------
T1 P1 TBS1
T1_INDX P1 TBS1_INDX
T1_INDX P2 TBS2_INDX
T1 P2 TBS2
T1_INDX P3 TBS3_INDX
T1 P3 TBS3
6 rows selected.
访问正常（索引亦导入成功）
18:12:07 [email protected] prod >col name for a50
18:12:19 [email protected] prod >r
1* select * from t1 where id=4
ID NAME
---------- --------------------------------------------------
4 C_OBJ#
4 TAB$
Elapsed: 00:00:00.00
Execution Plan
----------------------------------------------------------
Plan hash value: 1229066337
--------------------------------------------------------------------------------------------------------------
| Id | Operation | Name | Rows | Bytes | Cost (%CPU)| Time | Pstart| Pstop |
--------------------------------------------------------------------------------------------------------------
| 0 | SELECT STATEMENT | | 2 | 1030 | 1 (0)| 00:00:01 | | |
| 1 | PARTITION RANGE SINGLE | | 2 | 1030 | 1 (0)| 00:00:01 | 1 | 1 |
| 2 | TABLE ACCESS BY LOCAL INDEX ROWID| T1 | 2 | 1030 | 1 (0)| 00:00:01 | 1 | 1 |
|* 3 | INDEX RANGE SCAN | T1_INDX | 1 | | 1 (0)| 00:00:01 | 1 | 1 |
--------------------------------------------------------------------------------------------------------------
Predicate Information (identified by operation id):
---------------------------------------------------
3 - access("ID"=4)
Note
-----
- dynamic sampling used for this statement (level=2)
Statistics
----------------------------------------------------------
0 recursive calls
0 db block gets
5 consistent gets
0 physical reads
0 redo size
524 bytes sent via SQL*Net to client
419 bytes received via SQL*Net from client
2 SQL*Net roundtrips to/from client
0 sorts (memory)
0 sorts (disk)
2 rows processed
18:11:05 [email protected] prod >alter tablespace tbs1 read write;
Tablespace altered.
Elapsed: 00:00:02.10
18:14:34 [email protected] prod >alter tablespace tbs1_indx read write;
Tablespace altered.
三、备份载入的原介质
以上文档参考：谭怀远《让Oracle跑的更快2-基于海量数据的数据库设计和优化》，感谢作者！
欢迎大家访问我的个人网站 萌萌的IT人
时间： 2024-10-27 11:35:11