Thoughts -> Re: classiccmp server
On Mon, 2004-10-04 at 15:27 -0500, Dan Wright wrote:
> That brings me to another point, though -- if you had 2 seperate disks in the
> same enclosure, with no environmental isolation between them, you'd likely
> kill both disks with a head crash on either one.
Two stacks of 2.5" platters should fit in one 3.5" drive enclosure, and
give a slightly larger data surface as compared to a single 3.5" stack.
Platters will be thinner too so you can stack more of them; with two
head assemblies throughput should be improved; smaller diameter platters
mean less physical head movement and so faster seek times; lighter
platters mean less power for spindle motors and faster spin-up time;
3.5" enclosure means it fits into industry-standard enclosures...
I suspect reliability would be a key problem, coupled with heat
dissipation, and probably vibration problems from two drive motors. Not
to mention tooling/design costs...
:-)
> A head crash in a modern high-speed drive tends to throw a lot of
> pulverized platter and head bits around inside the disk, damaging the
> other platters that may not have been involved in the actual crash.
Modern head crashes are boring though. It's no good unless platters
physically explode :-) If only they could catch fire, too. There just
aren't enough modern-day computing war stories around...
> So I think you'd need to isolate the sections
> from each other with some kind of airtight (or at least
> really-small-bits-of-metal tight) seal; probably doable, but would take up
> more room that you're pretty low on already.
Don't modern drives embed servo information across platters though? So
as soon as you lose one platter the drive's dead anyway. A crash will
also likely introduce unacceptable drag into the head assembly, and
what's the point in having a drive running at less than 100%? To get
things back to an operational state you have to replace the whole lot
anyway.
I'm assuming you're kidding ;-)
cheers,
Jules
Received on Tue Oct 05 2004 - 05:33:45 BST
This archive was generated by hypermail 2.3.0
: Fri Oct 10 2014 - 23:37:21 BST