Hi JC, what is your peak redo generation rate per second?. . Redo would be flushed to logfile when any one of the following is met a) 1 MB of change b) every 3 seconds c) whenever there is a commit d) whenever log buffer is 1/3 full in your case, redo is already writing to the file anyways since any one of the conditions would be met and hence reducing the logbuffer would have no impact. you could test it though best regards sriram kumar On Sat, Apr 28, 2012 at 2:06 AM, CRISLER, JON A <JC1706@xxxxxxx> wrote: > Has anybody run into a case where the log buffer size was too big, and > negatively affected log file writes and log file sync waits ? > > I have a system that ignores my parameter for 32mb log file size and sets > it to 128mb, due to the granule size. The workaround is to lower the > granule size so that I can manually set the redo log buffer lower. But I > am not sure if this will have any positive affect on log file sync times. > Normal I/O for log file writes is fast at 2-4 ms, but log file syncs are > bad at 50 to 150 ms. > -- > //www.freelists.org/webpage/oracle-l > > > -- //www.freelists.org/webpage/oracle-l