I was wondering if there is something that could be done to extraction of bulk rows from the database.
Here is a quick recap.
I have datasources like SAP, Oracle, Sybase Siebel, flat files etc.,
I shall use the JDBC impl from the vendors in case of DAtabase and JCA Impl. for all others.
I am using JCA from JBoss.
When I extract from database, I set a prefetch buffer to extract 1000 rows or (whatever is scalable and fast to extract). But here I still have to iterate through large number of ResultSet.
I was wondering if we can set the rowsize and directly read data (Inputstream) which can be transfered in a temprovary buffer or something.
Basically JCA or JDBC for High speed extraction and Loading. Sicne this is just reading data and mostly Batch transfers, Is there a way, I can disable Transactions (Local and XA). I really need speed and nothing else. I do not want to be really in transactions, since I dont have to rollback or commit in a source database.
Any inputs or help ot suggestions would be appreciated.
This is probably a chance, if we could come out with a special request to optimize JDBC/JCA for high speed Extraction and Loading. (Reading and Writing large resultsets as a stream (chunkof data) ).
With respect to SAP, Siebel etc., I am not really concerned right now...