Weekly Classic Computer Trivia Question (20141205)

[Guy Sotomayor]

> On Dec 5, 2014, at 4:09 PM, Peter Corlett <abuse at cabal.org.uk> wrote:
> 
> On Fri, Dec 05, 2014 at 06:05:20PM -0500, Jerome H. Fine wrote:
> [...]
>> Assuming that 1 terabyte is 2 ** 40 bytes [...]
> 
> 

[snip]

> 
>    It is increasingly-rare to find exact capacities on data sheets, so I have
>    not checked against other manufacturers, but I gather that they have all
>    agreed to give their disks the same geometry for a given nominal advertised
>    capacity.
> 

The reason for that is primarily for spares.  Since the interfaces now in use are
LBA based, the drive when it detects a bad sector is free to reallocate the sector
from its spares pool.  This simplifies software in that it doesn't have to manage
bad blocks...the drive does it.  How much space is reserved on the drive is decided
by the manufacturer.  Usually there are some number of spare sectors allocated
to the cylinder itself and then there are entire spare cylinders.

The drive manufacturers don't agree on geometry.  You don't see it because it's
all hidden behind the LBA interface.

Drive capacity (and drive cost) is driven by 2 fundamental things: number of platters
and number of heads.  Yes, there are others such as quality of the media and motor,
but within a family the media and motor are common.

A drive manufacturer will "de-rate" a drive by not putting in all of the heads.  Actually 
a family of drives is 1, 2 (or sometimes 3) platters and 2n or 2n-1 heads (where n is 
the number of platters).

TTFN - Guy