Java/JDBC utilizing memory properly w/o thrashing

I've been given a very simple task of writing a process that will transfer data from one server/datbase to another server/database.  Seems easy.  The hard part that I'm having to deal with is the amount of data that I'm having to transfer (anywhere from a million to 50 million records).  I'm trying to design my process properly so that it takes into account memory resources when running, but I'm still getting a large amount of thrashing.  So my question is, how can I take this simple task and make it run optimally.

Here's a snippet of what I got so far.
..............
//grab 1st connection
Properties props = new java.util.Properties();
props.put("user", userid);
props.put("password", password);
props.put("block size", "512");
conn = DriverManager.getConnection(url, props);

conn.setTransactionIsolation(conn.TRANSACTION_NONE);
Statement st = conn.createStatement(ResultSet.TYPE_FORWARD_ONLY,ResultSet.CONCUR_READ_ONLY);
ResultSet rs = st.executeQuery("...")
...

//grab 2nd connection
Properties props = new java.util.Properties();
props.put("user", userid);
props.put("password", password);
props.put("block size", "512");
conn = DriverManager.getConnection(url, props);

//populate
conn.setAutoCommit(false);
PreparedStatement ps = conn.prepareStatement("...");
while (rs.next())
{
 ...
 ps.addBatch();
}
ps.executeBatch();
conn.commit();
conn.setAutoCommit(false);

//close connections and done!