If your data block corruptions are due to the asynchronous I/O bugs that have affected some versions of AIX, then you can avoid them by using raw datafiles. You may get a performance boost too!

I have a question regarding asynchronous I/O. When we enable disk_asynch_io at the Oracle level with asynchronous I/O enabled at the AIX level, block corruption takes place even though we set db_writer_process to 1. Can you suggest something why is it happening so?