使用在线重定义建分区表

Applies to:

Oracle Server - Enterprise Edition - Version 9.2.0.4 and later
Information in this document applies to any platform.
"Checked for relevance on 29-Sep-2010"

Goal

The purpose of this document is to provide step by step instructions on how to convert unpartitioned table to partitioned one using dbms_redefinition package.

Fix

1) Create unpartitioned table with the name unpar_table

SQL> CREATE TABLE unpar_table ( 
id NUMBER(10), 
create_date DATE, 
name VARCHAR2(100) 
);

2) Apply some constraints to the table:

SQL> ALTER TABLE unpar_table ADD ( 
CONSTRAINT unpar_table_pk PRIMARY KEY (id) 
); 

SQL> CREATE INDEX create_date_ind ON unpar_table(create_date);

3) Gather statistics on the table:

SQL> EXEC DBMS_STATS.gather_table_stats(USER, 'unpar_table', cascade => TRUE);

4) Create a Partitioned Interim Table:

SQL> CREATE TABLE par_table ( 
id NUMBER(10), 
create_date DATE, 
name VARCHAR2(100) 
) 
PARTITION BY RANGE (create_date) 
(PARTITION unpar_table_2005 VALUES LESS THAN (TO_DATE('01/01/2005', 'DD/MM/YYYY')), 
PARTITION unpar_table_2006 VALUES LESS THAN (TO_DATE('01/01/2006', 'DD/MM/YYYY')), 
PARTITION unpar_table_2007 VALUES LESS THAN (MAXVALUE));

5) Start the Redefinition Process:

a) Check the redefinition is possible using the following command:

SQL> EXEC Dbms_Redefinition.can_redef_table(USER, 'unpar_table');

b)If no errors are reported, start the redefintion using the following command:

SQL> BEGIN 
DBMS_REDEFINITION.start_redef_table( 
uname => USER,  
orig_table => 'unpar_table', 
int_table => 'par_table'); 
END; 
/

Note: This operation can take quite some time to complete.

c) Optionally synchronize new table with interim name before index creation:

SQL> BEGIN 
dbms_redefinition.sync_interim_table( 
uname => USER,  
orig_table => 'unpar_table', 
int_table => 'par_table'); 
END; 
/

d) Create Constraints and Indexes:

SQL> ALTER TABLE par_table ADD ( 
CONSTRAINT unpar_table_pk2 PRIMARY KEY (id) 
); 

SQL> CREATE INDEX create_date_ind2 ON par_table(create_date);

e) Gather statistics on the new table:

SQL> EXEC DBMS_STATS.gather_table_stats(USER, 'par_table', cascade => TRUE);

f) Complete the Redefintion Process:

SQL> BEGIN 
dbms_redefinition.finish_redef_table( 
uname => USER, 
orig_table => 'unpar_table', 
int_table => 'par_table'); 
END; 
/

At this point the interim table has become the "real" table and their names have been switched in the name dictionary. 

g) Remove original table which now has the name of the interim table:

SQL> DROP TABLE par_table; 

h)Rename all the constraints and indexes to match the original names.

ALTER TABLE unpar_table RENAME CONSTRAINT unpar_table_pk2 TO unpar_table_pk; 
ALTER INDEX create_date_ind2 RENAME TO create_date_ind;

i) Check whether partitioning is successful or not:

SQL> SELECT partitioned 
FROM user_tables 
WHERE table_name = 'unpar_table'; 

PAR 
--- 
YES 

1 row selected. 

SQL> SELECT partition_name 
FROM user_tab_partitions 
WHERE table_name = 'unpar_table'; 

PARTITION_NAME 
------------------------------ 
unpar_table_2005 
unpar_table_2006 
unpar_table_2007 

3 rows selected.

Please not that the 9i redefinition procedures has some restrictions: 

* There must be enough space to hold two copies of the table. 
* Primary key columns cannot be modified. 
* Tables must have primary keys. 
* Redefinition must be done within the same schema. 
* New columns added cannot be made NOT NULL until after the redefinition operation. 
* Tables cannot contain LONGs, BFILEs or User Defined Types. 
* Clustered tables cannot be redefined. 
* Tables in the SYS or SYSTEM schema cannot be redefined. 
* Tables with materialized view logs or materialized views defined on them cannot be redefined. 
* Horizontal sub setting of data cannot be performed during the redefinition.

来自 “ ITPUB博客 ” ，链接：http://blog.itpub.net/22531473/viewspace-747009/，如需转载，请注明出处，否则将追究法律责任。

转载于:http://blog.itpub.net/22531473/viewspace-747009/