I checked some ratios on our databases today and how these 2 issues:
For log space requests we had a value of 1001
Oracle suggests it be around 0
And gets to misses ratio for redo copy latch contention was almost 100% (97.8%)
Oracle sugesst less than or equal to 1%
Anyone have any suggestions on what I should do to improve these numbers.
For the latch contention I did check the log_simultaneous_copies paramter and found it equals the number of CPU's we have, should I add more.
Any suggestions would be appreciated
Don't be afraid to try something new. Amateurs built the Ark, professionals built the Titanic
It sounds to me like you may have inadequately sized logs and/or an inadequate number of log files. How are your redo logs configured? What size are they? How often are you doing a log switch?
Senior Database Administrator
Increase the size of the redo log buffer to reduce 'log space request'.
Increase the SHARED_POOL_SIZE to reduce latch contention.
A large number of Space request for log buffer is due to very small log buffer size. How ever in your case it is no high. Try to increase the log buffer size and bounce the database and monitor the performance.
I've been monitoring my log activity and my redo log space requests steadily increase each day, however it looks like the number only increases at log switches. Currently my logs switch about once every 2 hours. The performance really isn't an issue since it's a small development database, I'm just trying to learn what I can.
I've increased my log buffer a few times just to see the effect, my log buffer is currently 2097152, or 2048K and the requests still hit me at log switches.
Is it common to have the requests (usually 1-3) during log switches or is there something else I can do to help eliminate them?
If you have configured mirrored redo log files on the RAID system, you will experience this space request during the log switch.
Configure the redo log files in such a way that one member in RAID and the other in non mirrored disk.
In short, do not go for hardware level mirroring and software level mirroring for redo log files.
I have oracle mirrored logs on seperate disks, but on a non-raid system. There is only one controller on the machine. I suppose that is where the requests come from?
Click Here to Expand Forum to Full Width