mkfs.xfs options for DAC960 RAID volumes (long)

["Kai Leibrandt" <k_leibrandt@xxxxxxxxxxx>]

Hi all,
I just have a small (?) question regarding the following extract from man
mkfs.xfs, about the -d options:
"The sunit suboption is used to specify the stripe unit for a RAID device or
a logical volume. The suboption value has to be specified in 512­byte block
units. This suboption ensures that data allocations will be stripe unit
aligned when the current end of file is being extended and the file size is
larger than 512KB. Also inode allocations and the internal log will be
stripe unit aligned.
The swidth suboption is used to specify the stripe width for a RAID device
or a striped logical volume. The suboption value has to be specified in
512­byte block units. This suboption is required if -d sunit has been
specified and it has to be a multiple of the -d sunit suboption. The stripe
width will be the preferred iosize returned in the stat(2) system call."

My question is about the block sizes in particular. As my DAC960 has a
segment size of 8k and a stripe size of 64, it seems to me as though these
valuse should be matched by the sunit and swidth options (I'm assuming these
will align io transfers to the respective sizes). In the man page it says
that these are to be specified in 512byte units, so I'm guessing that I
should use -d sunit=16,swidth=128 to achieve vaues of 8Kbyte and 64Kbyte
respectively, but the output of mkfs and xfs_info then tell me sunit is then
actually 2 and swidth=16blks... Is the blocking size really 512byte as
stated in the man page or is it actually the size of a page (i.e. 4Kb with
ia32 linux)? In that case it would suddenly all fit - 16 blocks of 4Kb make
64Kb stripe size...

So what values of sunit and swidth should be used for 8k segment /64k stripe
size ahrdware RAID volumes on ia32?

Tomorrow I'll be running some bonnie benchmarks to see what the difference
may be... many thanks for any info!

Kai.