Our database is running on hpux on which AIO is not possible to
achieve(except use raw device which is out of our considertion). And the "log
file parallel write" is very high. Since we use two redo groups to mutliplex
the redo files, and the files can only be written sequentially. So I figure
out an idea to use one group of online redo log which are mirrored. If it is
mirrored, the redo log will be written in parallel which can reduce the wait
time.
Do you think this method is feasible and what's the disadvanatge?
--
http://www.freelists.org/webpage/oracle-l
Received on Tue Oct 02 2007 - 19:25:24 CDT
