RE: I/O and db_file_multiblock_read_count

I have found that I/O requests that involve the same spindles more than
onec in a single
request simply do not perform. It is generally better to hit the LUN
more times for I/Os
sized rightly to not wrap an entire stripe set.
 
So, if I have 2 disks and intend to scan them with 1MB reads, uh, I
don't set the stripe
width to 32KB. Having the array line up 16 chucks on each disk to
satisfy a single
transfer request is just a mess.
 
In fact, I generally set stripe width to a value larger than 1MB if the
kit supports it.
 
YMMV ... this is a religious topic you know... :-)
 
 
 


________________________________

        From: Vlad Sadilovskiy [mailto:vlovsky@xxxxxxxxx] 
        Sent: Monday, December 11, 2006 1:29 PM
        To: kevinc@xxxxxxxxxxxxx
        Cc: oracle-l
        Subject: Re: I/O and db_file_multiblock_read_count
        
        
        How does number of disk hits for larger I/O corresponds to
number of hits for smaller I/O in respect to the whole process. It
should not be more should it? Unless same data gets read twice.

Other related posts: