Well, I finally bothered to install the array managment software for my
DPT card (WinNT as OS, sorry), and noticed that it actually tracked some
useful statistics. This particular server is a "forms server" for
Oracle 805, whatever that is, so I created the RAID array with the
default 32K blocks (default for the DPT card, that is). Conviently
enough, the software tracks the reads and writes of each size. I.E.
<1K, from 1K to 2K, from 2K to 4K, from 4K to 8K, from 8K to 16K, 16K to
32K, etc. It seems to me that with this information, tracked over a
reasonable period of time of "normal use" of the server would allow you
to choose the most appropraite block size to use when creating a RAID
array. Now for the question. I know that the proc filesystem collects
these statistics for the individual SCSI disks, but I haven't been able
to find if or where it collects these statistics for the RAID devices.
So, does it collect them, and if so, where? If not, why?
Greg
