11.2.0.3.3 impdp compress table

Question

HI ML :
源库 ： 10.2.0.3 compress table
target ： 11.2.0.3.3 impdp 源端的compress tables，在目标端是否是compress table
之前在10g库直接 通过impdp dblink 导入时候 发现，入库的表需要手工做move compress。
MOS 文档给的的测试时 在10g开始 支持导入自动维护compress table ：
Oracle Server - Enterprise Edition - Version: 9.2.0.1 to 11.2.0.1 - Release: 9.2 to 11.2
Information in this document applies to any platform.
Symptoms
Original import utility bypasses the table compression or does not compress, if the table is precreated as compressed. Please follow the next example that demonstrates this.
connect / as sysdba
create tablespace tbs_compress datafile '/tmp/tbs_compress01.dbf' size 100m;
create user test identified by test default tablespace tbs_compress temporary tablespace temp;
grant connect, resource to test;
connect test/test
-- create compressed table
create table compressed
id number,
text varchar2(100)
) pctfree 0 pctused 90 compress;
-- create non-compressed table
create table noncompressed
id number,
text varchar2(100)
) pctfree 0 pctused 90 nocompress;
-- populate compressed table with data
begin
for i in 1..100000 loop
insert into compressed values (1, lpad ('1', 100, '0'));
end loop;
commit;
end;
-- populate non-compressed table with identical data
begin
for i in 1..100000 loop
insert into noncompressed values (1, lpad ('1', 100, '0'));
end loop;
commit;
end;
-- compress the table COMPRESSED (previous insert doesn't use the compression)
alter table compressed move compress;
Let's now take a look at data dictionary to see the differences between the two tables:
connect test/test
select dbms_metadata.get_ddl ('TABLE', 'COMPRESSED') from dual;
DBMS_METADATA.GET_DDL('TABLE','COMPRESSED')
CREATE TABLE "TEST"."COMPRESSED"
( "ID" NUMBER,
"TEXT" VARCHAR2(100)
) PCTFREE 0 PCTUSED 90 INITRANS 1 MAXTRANS 255 COMPRESS LOGGING
STORAGE(INITIAL 65536 NEXT 1048576 MINEXTENTS 1 MAXEXTENTS 2147483645
PCTINCREASE 0 FREELISTS 1 FREELIST GROUPS 1 BUFFER_POOL DEFAULT)
TABLESPACE "TBS_COMPRESS"
1 row selected.
SQL> select dbms_metadata.get_ddl ('TABLE', 'NONCOMPRESSED') from dual;
DBMS_METADATA.GET_DDL('TABLE','NONCOMPRESSED')
CREATE TABLE "TEST"."NONCOMPRESSED"
( "ID" NUMBER,
"TEXT" VARCHAR2(100)
) PCTFREE 0 PCTUSED 90 INITRANS 1 MAXTRANS 255 NOCOMPRESS LOGGING
STORAGE(INITIAL 65536 NEXT 1048576 MINEXTENTS 1 MAXEXTENTS 2147483645
PCTINCREASE 0 FREELISTS 1 FREELIST GROUPS 1 BUFFER_POOL DEFAULT)
TABLESPACE "TBS_COMPRESS"
1 row selected.
col segment_name format a30
select segment_name, bytes, extents, blocks from user_segments;
SEGMENT_NAME BYTES EXTENTS BLOCKS
COMPRESSED 2097152 17 256
NONCOMPRESSED 11534336 26 1408
2 rows selected.
The table COMPRESSED needs fewer storage space than the table NONCOMPRESSED. Now, let's export the tables using the original export utility:
#> exp test/test file=test_compress.dmp tables=compressed,noncompressed compress=n
About to export specified tables via Conventional Path ...
. . exporting table COMPRESSED 100000 rows exported
. . exporting table NONCOMPRESSED 100000 rows exported
Export terminated successfully without warnings.
and then import them back:
connect test/test
drop table compressed;
drop table noncompressed;
#> imp test/test file=test_compress.dmp tables=compressed,noncompressed
. importing TEST's objects into TEST
. . importing table "COMPRESSED" 100000 rows imported
. . importing table "NONCOMPRESSED" 100000 rows imported
Import terminated successfully without warnings.
Verify the extents after original import:
col segment_name format a30
select segment_name, bytes, extents, blocks from user_segments;
SEGMENT_NAME BYTES EXTENTS BLOCKS
COMPRESSED 11534336 26 1408
NONCOMPRESSED 11534336 26 1408
2 rows selected.
=> The table compression is gone.
Cause
This is an expected behaviour. Import is not performing a bulk load/direct path operations, so the data is not inserted as compressed.
Only Direct path operations such as CTAS (Create Table As Select), SQL*Loader Direct Path will compress data. These operations include:
•Direct path SQL*Loader
•CREATE TABLE and AS SELECT statements
•Parallel INSERT (or serial INSERT with an APPEND hint) statements
Solution
The way to compress data after it is inserted via a non-direct operation is to move the table and compress the data:
alter table compressed move compress;
Beginning with Oracle version 10g, DataPump utilities (expdp/impdp) perform direct path operations and so the table compression is maintained, like in the following example:
- after crating/populating the two tables, export them with:
#> expdp test/test directory=dpu dumpfile=test_compress.dmp tables=compressed,noncompressed
Processing object type TABLE_EXPORT/TABLE/TABLE
. . exported "TEST"."NONCOMPRESSED" 10.30 MB 100000 rows
. . exported "TEST"."COMPRESSED" 10.30 MB 100000 rows
Master table "TEST"."SYS_EXPORT_TABLE_01" successfully loaded/unloaded
and re-import after deletion with:
#> impdp test/test directory=dpu dumpfile=test_compress.dmp tables=compressed,noncompressed
Processing object type TABLE_EXPORT/TABLE/TABLE_DATA
. . imported "TEST"."NONCOMPRESSED" 10.30 MB 100000 rows
. . imported "TEST"."COMPRESSED" 10.30 MB 100000 rows
Job "TEST"."SYS_IMPORT_TABLE_01" successfully completed at 12:47:51
Verify the extents after DataPump import:
col segment_name format a30
select segment_name, bytes, extents, blocks from user_segments;
SEGMENT_NAME BYTES EXTENTS BLOCKS
COMPRESSED 2097152 17 256
NONCOMPRESSED 11534336 26 1408
2 rows selected.
=> The table compression is kept.
===========================================================
1 到底11.2.0.3 是否 支持impdp自动维护compress table 通过dblink 方式？
2
This is an expected behaviour. Import is not performing a bulk load/direct path operations, so the data is not inserted as compressed.
Only Direct path operations such as CTAS (Create Table As Select), SQL*Loader Direct Path will compress data. These operations include:
•Direct path SQL*Loader
•CREATE TABLE and AS SELECT statements
•Parallel INSERT (or serial INSERT with an APPEND hint) statements
Solution
The way to compress data after it is inserted via a non-direct operation is to move the table and compress the data:
以上意思在10g之前是 必须以上方式才能支持目标端入库压缩表，10g开始支持自动压缩？ 貌似 10g也需要手工move。

Answer 1

ODM TEST:
Oracle Database 11g Enterprise Edition Release 11.2.0.3.0 - 64bit Production
With the Partitioning, OLAP, Data Mining and Real Application Testing options
SQL> create table nocompres tablespace users as select * from dba_objects;
Table created.
SQL> create table compres_tab tablespace users as select * from dba_objects;
Table created.
SQL> alter table compres_tab compress 3;
Table altered.
SQL> alter table compres_tab move ;
Table altered.
select bytes/1024/1024 ,segment_name from user_segments where segment_name like '%COMPRES%'
BYTES/1024/1024 SEGMENT_NAME
              3 COMPRES_TAB
              9 NOCOMPRES
C:\Users\ML>expdp  maclean/oracle dumpfile=temp:COMPRES_TAB2.dmp  tables=COMPRES_TAB
Export: Release 11.2.0.3.0 - Production on Fri Sep 14 12:01:12 2012
Copyright (c) 1982, 2011, Oracle and/or its affiliates.  All rights reserved.
Connected to: Oracle Database 11g Enterprise Edition Release 11.2.0.3.0 - 64bit Production
With the Partitioning, OLAP, Data Mining and Real Application Testing options
Starting "MACLEAN"."SYS_EXPORT_TABLE_01":  maclean/******** dumpfile=temp:COMPRES_TAB2.dmp tables=COMPRES_TAB
Estimate in progress using BLOCKS method...
Processing object type TABLE_EXPORT/TABLE/TABLE_DATA
Total estimation using BLOCKS method: 3 MB
Processing object type TABLE_EXPORT/TABLE/TABLE
. . exported "MACLEAN"."COMPRES_TAB"                     7.276 MB   75264 rows
Master table "MACLEAN"."SYS_EXPORT_TABLE_01" successfully loaded/unloaded
Dump file set for MACLEAN.SYS_EXPORT_TABLE_01 is:
  D:\COMPRES_TAB2.DMP
Job "MACLEAN"."SYS_EXPORT_TABLE_01" successfully completed at 12:01:20
C:\Users\ML>impdp maclean/oracle remap_schema=maclean:maclean1 dumpfile=temp:COMPRES_TAB2.dmp
Import: Release 11.2.0.3.0 - Production on Fri Sep 14 12:01:47 2012
Copyright (c) 1982, 2011, Oracle and/or its affiliates.  All rights reserved.
Connected to: Oracle Database 11g Enterprise Edition Release 11.2.0.3.0 - 64bit Production
With the Partitioning, OLAP, Data Mining and Real Application Testing options
Master table "MACLEAN"."SYS_IMPORT_FULL_01" successfully loaded/unloaded
Starting "MACLEAN"."SYS_IMPORT_FULL_01":  maclean/******** remap_schema=maclean:maclean1 dumpfile=temp:COMPRES_TAB2.dmp
Processing object type TABLE_EXPORT/TABLE/TABLE
Processing object type TABLE_EXPORT/TABLE/TABLE_DATA
. . imported "MACLEAN1"."COMPRES_TAB"                    7.276 MB   75264 rows
Job "MACLEAN"."SYS_IMPORT_FULL_01" successfully completed at 12:01:50
  1* select bytes/1024/1024 ,segment_name from user_segments where segment_name like '%COMPRES%'
SQL> /
BYTES/1024/1024 SEGMENT_NAME
              3 COMPRES_TAB
SQL> drop table compres_tab;
Table dropped.
C:\Users\ML>exp maclean/oracle tables=COMPRES_TAB file=compres1.dmp
Export: Release 11.2.0.3.0 - Production on Fri Sep 14 12:03:19 2012
Copyright (c) 1982, 2011, Oracle and/or its affiliates.  All rights reserved.
Connected to: Oracle Database 11g Enterprise Edition Release 11.2.0.3.0 - 64bit Production
With the Partitioning, OLAP, Data Mining and Real Application Testing options
Export done in ZHS16GBK character set and AL16UTF16 NCHAR character set
About to export specified tables via Conventional Path ...
. . exporting table                    COMPRES_TAB      75264 rows exported
Export terminated successfully without warnings.
C:\Users\ML>
C:\Users\ML>imp maclean/oracle  fromuser=maclean touser=maclean1  file=compres1.dmp
Import: Release 11.2.0.3.0 - Production on Fri Sep 14 12:03:45 2012
Copyright (c) 1982, 2011, Oracle and/or its affiliates.  All rights reserved.
Connected to: Oracle Database 11g Enterprise Edition Release 11.2.0.3.0 - 64bit Production
With the Partitioning, OLAP, Data Mining and Real Application Testing options
Export file created by EXPORT:V11.02.00 via conventional path
import done in ZHS16GBK character set and AL16UTF16 NCHAR character set
. importing MACLEAN's objects into MACLEAN1
. . importing table                  "COMPRES_TAB"      75264 rows imported
Import terminated successfully without warnings.
SQL> conn maclean1/oracle
Connected.
  1* select bytes/1024/1024 ,segment_name from user_segments where segment_name like '%COMPRES%'
SQL> /
BYTES/1024/1024 SEGMENT_NAME
              8 COMPRES_TAB
                 我的理解 对于direct load 总是能保持compression
但是 imp默认是conventional path 即使用普通的走buffer cache的INSERT实现导入 所以无法保持compression
而impdp不管 access_method 是external_table还是direct_path 模式都可以保持compression