The db file parallel write
wait event occurs when the process,
typically DBWR, has issued multiple
I/O requests in parallel to write
dirty blocks from the buffer cache
to disk, and is waiting for all
requests to complete. If this is a
significant portion of the total
wait time it is not necessarily
having a large impact on on user
sessions. However this combined with
the waits from free buffer waits
shows that this is impacting user
sessions. this
Recommendation: Be sure async I/O is enabled for the platform, because you want to be absolutely sure that the maximum possible bandwidth is available to DBWR, also make sure the disk layout allows the most amount of performance to the disks RAID 1+0). Consideration should be given to converting to RAW for better performance along with turning on ASYNC I/O. Also ensure there is enough buffer cache to prevent thrashing going on with the buffer cache.
