I'm evaluating JBoss and, one of the first things I made is to do two tests, writing and reading big amounts of data. (Jboss 4.0.2 and Oracle)
- Writing a few thousands of rows to a simple table is very slow. I have a client, calls a simple stateless session bean, and it sends the data using RMI-IIOP. Client side finishes its job pretty quickly, but as I observe JBoss console I see only a few records written to database per second, and it goes like that for almost an hour.
- Reading 10.000 records (using a simple findAll EJB-QL) from a simple table (only one relation) takes about 10 seconds. I think this is unacceptably slow. I have checked out some documents and tried a few read-ahead parameters but it only improved a few seconds.
Is there something I'm doing wrong? What is the best way to read a simple, yet very crowded (tens of thousands of records) tables?
Do you have biDir or uniDir CMR ? Using findAll query on lot of table rows is not the bets way how to retrive huge data from DB.