Re: linux memory settings on OL 6.2 ( red hat compatible )

>> echo deadline >/sys/block/sdb/queue/scheduler
>>
>> Substitute your own devices for "sdb".

I am not sure, but it will not harm anything. IO scheduler lives between the generic block driver and the device driver itself. However, the only time IO scheduler kicks in is when you have an I/O queue. If iostat shows no queueing on any of the devices, then you will see no benefits. The function of IO scheduler is to prioritize the I/O requests waiting for the device. If there is no waiting for IO, then it doesn't matter what you set the scheduler to.
Also, I know that EMC has IO scheduler on the SAN side, at least for Symmetrix series. I am not sure about Carrion (or is it Clariion? I am a foreigner and I frequently make spelling mistakes) but any half way intelligent SAN capable of using FC disks has an IO scheduler.

>
>
> # With such a large memory, you don't want to throw out inodes and
> directory entries quickly. I would set something like
> vm.vfs_cache_pressure = 20 to retain them as long as possible.
>
> Well we don't have anything on any file systems really ... everything
> cached in database buffer cache.

Except, of course, executables. Those inodes are the ones that Linux will be throwing out of buffer cache if you increase cache pressure. That means that when you invoke sqlplus, the inode is unlikely to be in the buffer cache and will have to be re-read from disk. That, of course, apply to everything else, from cron to vi editor.

-- 
Mladen Gogala
The Oracle Whisperer
http://mgogala.byethost5.com