>Our UNIX SA says " Cooked Filesystems have their own Asynchronous Mechanisim that equally replaces Oracle Async IO" ...Uh, tell him to keep his day job. Having buffers copied into kernel buffers from the SGA so that bdflush can eventually get to them is in no way a tradeoff for Oracle making async I/O calls. Not to mention, all the important writes (dbwr,lgwr) are synced (either with open(,O_SYNC,) or fsync so the topic is nonsensical. -- //www.freelists.org/webpage/oracle-l