DBMS_DATAPUMP parallel parameter did not work

Question

Hi, I am using DBMS_DATADUMP with network link to load large table directly from one database to another database. I used parallel parameter to improve the performance. But looks the parallel did not work. Still only one worker to handle the import.
The status is
Job: LOADING
Owner: TRI
Operation: IMPORT
Creator Privs: TRUE
GUID: 25DB4B2BE420406B82A7AE159CF1E626
Start Time: Wednesday, 22 April, 2009 15:37:04
Mode: TABLE
Instance: orcl
Max Parallelism: 4
EXPORT Job Parameters:
IMPORT Job Parameters:
Parameter Name Parameter Value:
INCLUDE_METADATA 0
TABLE_EXISTS_ACTION TRUNCATE
State: EXECUTING
Bytes Processed: 0
Current Parallelism: 4
Job Error Count: 0
Worker 1 Status:
Process Name: DW01
State: EXECUTING
Object Schema: TRI
Object Name: ORDER_LINES
Object Type: TABLE_EXPORT/TABLE/TABLE_DATA
Completed Objects: 1
Total Objects: 1
Worker Parallelism: 1
Worker 2 Status:
Process Name: DW02
State: WORK WAITING
My source database is 10.2.0.4 compatible=10.2.0
target database is 11.1.0.6
My table is around 3G
The API I am using is
my_handle := dbms_datapump.open(operation => 'IMPORT',job_mode => 'TABLE',
remote_link => my_db_link, job_name => my_job_name ,version=>'LATEST' ) ;
dbms_datapump.set_parameter (my_handle, 'TABLE_EXISTS_ACTION','TRUNCATE');
dbms_datapump.set_parameter (my_handle, 'INCLUDE_METADATA',0);
dbms_datapump.metadata_filter (handle => my_handle, name => 'SCHEMA_EXPR',
value=>'IN (''ORDER'')');
dbms_datapump.metadata_filter (handle => my_handle, name => 'NAME_LIST',
value => '''ORDER_LINES''');
dbms_datapump.metadata_remap(my_handle,'REMAP_SCHEMA',old_value=>'ORDER',value=>'TRI');
dbms_datapump.set_parallel(my_handle,16);
dbms_datapump.start_job(my_handle);
Edited by: tonym on Apr 23, 2009 10:49 AM

Answer 1

Then I test to use API and network link to export large table from remote database with parallel parameter.
my_handle := dbms_datapump.open(operation => 'EXPORT',job_mode => 'TABLE',remote_link => my_db_link, job_name => my_job_name ,version=>'LATEST' ) ;
DBMS_DATAPUMP.add_file( handle => my_handle, filename => 'test1.dmp', directory => 'DATA_PUMP_DIR');
DBMS_DATAPUMP.add_file( handle => my_handle, filename => 'test2.dmp', directory => 'DATA_PUMP_DIR');
DBMS_DATAPUMP.add_file( handle => my_handle, filename => 'test3.dmp', directory => 'DATA_PUMP_DIR');
DBMS_DATAPUMP.add_file( handle => my_handle, filename => 'test4.dmp', directory => 'DATA_PUMP_DIR');
dbms_datapump.metadata_filter (handle => my_handle, name => 'SCHEMA_EXPR',value=>'IN (''INVENTORY'')');
dbms_datapump.metadata_filter (handle => my_handle, name => 'NAME_LIST',value => '''INV_TRANSACTIONS''');
dbms_datapump.set_parallel(my_handle,4);
dbms_datapump.start_job(my_handle);
Looks it did not use parallel either. This table is around 3G too.
status:
Job: LOADING_2
Operation: EXPORT
Mode: TABLE
State: EXECUTING
Bytes Processed: 0
Current Parallelism: 4
Job Error Count: 0
Dump File: C:\ORACLE\ADMIN\OW\DPDUMP\TEST1.DMP
bytes written: 69,632
Dump File: C:\ORACLE\ADMIN\OW\DPDUMP\TEST2.DMP
bytes written: 4,096
Dump File: C:\ORACLE\ADMIN\OW\DPDUMP\TEST3.DMP
bytes written: 4,096
Dump File: C:\ORACLE\ADMIN\OW\DPDUMP\TEST4.DMP
bytes written: 4,096
Worker 1 Status:
Process Name: DW04
State: WORK WAITING
Worker 2 Status:
Process Name: DW05
State: EXECUTING
Object Schema: INVENTORY
Object Name: INV_TRANSACTIONS
Object Type: TABLE_EXPORT/TABLE/TABLE_DATA
Completed Objects: 1
Total Objects: 1
Completed Rows: 4,652,330
Worker Parallelism: 1
Edited by: tonym on Apr 23, 2009 10:53 AM