data switchs su,sw and sunit,swidth

data switchs su,sw and sunit,swidth

[Linda Walsh]

Resending this as I never saw it show up on the list  (sent out
yesterday) (while other the messages came back in under 15 minutes 
or so)...

Have started to use RAID on a few of my disks and forgot about the
xfs 'su*sw' and 'sunit*swidth' options.

>From what I get in reading the manpage, 'su' is used with 'sw' and 
'sunit' is used with 'swidth'?

The RAID controller in one of my machines uses a "strip element" size, 
expressed in bytes, allowed values seem to be limited to powers 
of 512*2^*[1..11] (512B up to 1MB) (though as previously noticed, although
xfs's manpages claims to allow one expresses sizes with the unit 'm', it
only permits .25m (256k),  I guess I never tried seeing if the command line 
would take floating point ;^) ).

I believe 'su' would be set to the 'strip element size' (in k or m).

Then, for RAID 1 (mirror) would 'sw'==1?  Would setting the  su/sw  value
for a RAID 1 actually matter in any way?  Ie, technically -- it would 
fill in numbers for OS book-keeping, but wouldn't change anything in 
terms of performance or layout, vs. 'physically' -- where it could change
disk layout or performance?

At RAID 0, I'd guess  sw==2?

In RAID 5, would it be   sw  == #Disks-1?  So even w/6 disks, it still only
uses 1 disk for parity and  sw == 5?  

I wonder what becomes a max-safe RAID 5 size? (or is the number of parity
disks a settable option with RAID 5?)

Thanks!...
Linda



_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

Re: data switchs su,sw and sunit,swidth

[Michael Monnerie]

[-- Attachment #1.1: Type: text/plain, Size: 1631 bytes --]

On Sonntag 19 Juli 2009 Linda Walsh wrote:
> Then, for RAID 1 (mirror) would 'sw'==1?  Would setting the  su/sw
>  value for a RAID 1 actually matter in any way?  Ie, technically --
> it would fill in numbers for OS book-keeping, but wouldn't change
> anything in terms of performance or layout, vs. 'physically' -- where
> it could change disk layout or performance?
>
> At RAID 0, I'd guess  sw==2?

Both RAID0 and RAID1 use sw=0.

> In RAID 5, would it be   sw  == #Disks-1?  So even w/6 disks, it
> still only uses 1 disk for parity and  sw == 5?

RAID5: sw = #Disks-1 ( so with 8 disks use 7)
RAID6: sw = #Disks-2 ( so with 8 disks use 6)

> I wonder what becomes a max-safe RAID 5 size? (or is the number of
> parity disks a settable option with RAID 5?)

RAID5 always only has 1 parity disk (well, technically it's not a 
physical disk, but the parity is distributed over all disks in the 
array).
Don't know what you mean by "max-safe" size. The more disks you have, 
the bigger the chance that a single disk breaks. Also, I tested with 
Areca controllers, using more than 7 disks in a single RAID array 
doesn't improve speed anymore. So I use RAID-6 for up to 8 disks, and 
make it RAID-60 for up to 16 disks (with Areca controllers).

mfg zmi
-- 
// Michael Monnerie, Ing.BSc    -----      http://it-management.at
// Tel: 0660 / 415 65 31                      .network.your.ideas.
// PGP Key:         "curl -s http://zmi.at/zmi.asc | gpg --import"
// Fingerprint: AC19 F9D5 36ED CD8A EF38  500E CE14 91F7 1C12 09B4
// Keyserver: wwwkeys.eu.pgp.net                  Key-ID: 1C1209B4


[-- Attachment #1.2: This is a digitally signed message part. --]
[-- Type: application/pgp-signature, Size: 197 bytes --]

[-- Attachment #2: Type: text/plain, Size: 121 bytes --]

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs