Re: What is a disk arm?

[Charles Wilt <charles.wilt@xxxxxxxxx>]

Cool...actually I think the disk vs. arms is somewhat back into play
now-a-days with SANs..

Consider a SAN of 100 disks, and you want to provide your i with 10 disks.

You could simply pick 10 individual disks.  Or you could take a 1/10 disk
slice across all 100 physical disks.  Assuming your i is the only thing
using the SAN, you would theoretically have 100 arms but only 10 disks your
IOPs would increase 10-fold!

Again, as I understand it...properly slicing up a SANs physical disk is
_very_ important.

Charles



On Thu, Jun 13, 2013 at 8:18 PM, DrFranken <midrange@xxxxxxxxxxxx> wrote:

> Correct. The 9332 family of drives - models 400 and 600 were supported
> on AS/400 - had two sets of heads each covering 1/2 of the platters (or
> possibly 1/2 of the tracks on all the platters, can't be certain any
> longer. :-)  ).   As a consequence the system saw 2 arms per drive of
> 1/2 the capacity or 200/200 and 300/300 MB, yes MB.  When Mirroring and
> Checksum (precursor to RAID) came you had to be very careful how you
> picked your arms or you could have multiple components in the same drive
> and thus no protection!
>
>          - Larry "DrFranken" Bolhuis
>
> www.frankeni.com
> www.iDevCloud.com
> www.iInTheCloud.com
>
> On 6/13/2013 4:54 PM, Charles Wilt wrote:
>
> > As I understand it....
> >
> > Back before my time..disks had multiple independent arms..thus "disk
> arms"
> > vs. "disks"
> >
> > now-a-days, 1 disk = 1 disk arm
> >
> > Charles
> >
> >
> > On Thu, Jun 13, 2013 at 4:24 PM, Buck Calabro <kc2hiz@xxxxxxxxx> wrote:
> >
> > > It's often said that there's no stupid question, but I think I'm about
> > > to test this adage.  I've got my cable set between the UPS and the 720
> > > (8202-E4C).  I got the CPI0962 (UPS now attached) message.  I don't have
> > > a UPS power handling program written yet, so I want to set QUPSDLYTIM to
> > > a value that will allow the machine to write main store to disk before
> > > crashing.  I look at
> http://pic.dhe.ibm.com/infocenter/iseries/v7r1m0/index.jsp?topic=%2Frzahr%2Frzahrdlytimco.htm
> > > (Determining the value of QUPSDLYTIM) and I can see that the value is
> > > heavily dependent on the number of disk arms and main storage.
> > > WRKSYSSTS and a calculator tell me I have the 32GB we paid for :-)
> > > WRKDSKSTS tells me I have 8 198C disk units and 12 19B0 disk units.  If
> > > that's 20 arms, I'm hosed - the table says it'll take (worst case) 58
> > > minutes to write 32GB RAM to disk (!) [1]
> > >
> > > And so, I don't think those 20 disk units equal 20 disk arms but my
> > > Google-fu has failed me in finding a conversion table.  I tried Power
> > > 720, 8202-E4C, 198C, 19B0.  Is 'disk arm' an imaginary unit of measure
> > > like CPW?
> > >    --buck
> > >
> > > [1] These numbers seem crazy even if worst case.  It's really going to
> > > take an hour to write 32GB of data from RAM to disk?
> > > --
> > > This is the Midrange Systems Technical Discussion (MIDRANGE-L) mailing
> list
> > > To post a message email: MIDRANGE-L@xxxxxxxxxxxx
> > > To subscribe, unsubscribe, or change list options,
> > > visit: http://lists.midrange.com/mailman/listinfo/midrange-l
> > > or email: MIDRANGE-L-request@xxxxxxxxxxxx
> > > Before posting, please take a moment to review the archives
> > > at http://archive.midrange.com/midrange-l.
> --
> This is the Midrange Systems Technical Discussion (MIDRANGE-L) mailing list
> To post a message email: MIDRANGE-L@xxxxxxxxxxxx
> To subscribe, unsubscribe, or change list options,
> visit: http://lists.midrange.com/mailman/listinfo/midrange-l
> or email: MIDRANGE-L-request@xxxxxxxxxxxx
> Before posting, please take a moment to review the archives
> at http://archive.midrange.com/midrange-l.