Guna,
I had the same problem, when having to select large amount of data (10k+ rows having over 150 collumns) from JDE dbase to use in mapping.
The JDBC adapter had no chance pulling this amount of data to the IS, and I was getting OutOfMemory even before the JDBC adapter finished.
The workaround for this could be not to use JDBC adapters at all, and configure the incoming and outgoing connections in WmDB and then access them in java service.
This way you might be able to get the data as stream into the IS (if it’s blobs) or somehow optimize the ResultSet object returned, and have them stored on filesystem straight away or directly stream them into another java service that would perform the insert.
I have seen this working quite ok, but wM weren’t very happy of us using the WmDB connection in our own java code.
It is a pity, that Wm JDBC adapters don’t support streams as inputs/outputs.
Hope this helps
//Matt