[rescue] Small RAID array setup
Curtis H. Wilbar Jr.
rescue at hawkmountain.net
Fri Jun 27 10:07:15 CDT 2003
>From: Andrew Weiss <ajwdsp at cloud9.net>
>I had an idea a bit ago... why not just RAID disks...
>New HD's with two spindles in the drive that counter-rotate *(like the
>original drives from the 60's)
>The spindles mesh together... i.e. one platter on one... then platter
>on the other... then back
>and the rw heads are on the sides of the drive... one on each
>indicated by the pipe position.
>This drive can either be configured for fast access as one stripe or
>redundancy via internal mirroring.
>Then you can stripe this set of mirrors and be done with things.
One major flaw.... it is likely a drive failure will take out both
spindles (either electronics or pollution from a head crash, etc)
and your mirror is useless.
Also, if you want to do that your going to have to accept smaller
platter sizes for everything to fit in the same enclosure... which
reduce capacity. With the head placement your talking about you
could probably get an extra platter or two in to compensate though.
However, the expensive parts of the inside of a disk (as I understand
it) are platters and head assemblies. Your design would produce a
drive with more performance and some interesting flexibilities, but
would either lower or not provide any increase in space, all while
providing a significant increase in cost.
While an interesting idea, there is no economic gain for such a beast
(that I can see).
Your idea is a variation on an idea I had ages ago about putting a
second set of heads in a disk drive to improve random access time,
doing concurrent reads, mising reads/writes, etc. Problem with that
idea is it too increases costs without a significant return in the
Any idea if in any recent (within 5ish years) either of these ideas
(Andrew's or mine) has actually been put in a production drive
(magnetic, optical, etc) ?
>rescue list - http://www.sunhelp.org/mailman/listinfo/rescue
More information about the rescue