This content has been marked as final. Show 7 replies
Data Pump import using the NETWORK_LINK parameter will avoid the need for a file, but if you cannot "select * from dst_table@db_src" in the destination database, then data pump network import will not help, either, since it does a similar "select from source over database link" approach to move the data.
Data Pump import still performs INSERTs, so just switching to data pump may not solve a performance issue.
Data Pump does support what you want to do, but you need to have a db link created on the target pointing to the source. It seems like you have it the other way around. This is not supported for DataPump.
If you are on the target or dest database and have a db link pointing to the source db_link_source then use this command:
impdp user/password tables=userx.tablex network_link=db_link_source ...
This will import that table from the source into your local destination database.
I have to do this using DataPump API :)
NETWORK_LINK is always a pull so i think you have no chance unless you can create the db link the other way round. How about some other form of sync - streams for example?
Network link is supported in the Data Pump api. Like the post prior to this one says, network link is a pull, never a push, so it can be done, but the network link needs to be defined on the target/destination pointing to the source.
Thank you all for your advice.
Sorry Harry, but I don't know any other sync method. If DataPump needs the db link the other way around, I will stick to INSERT INTO dest_table@remote_link :(
Thank you again :)
thanks for all good info.. during NETWORK_LINK pull do we need to drop destination table before pulling from source?
or we use table_exist= replace option?
all objects connected to table will also get import?
Edited by: khallas301 on Dec 13, 2012 11:40 AM