[plug] linking to different drives

[John Summerfield]

> On Thu, 21 Oct 1999, Matt Kemner wrote:
> 
> > It is reserved for root - so when the drive gets close to running out,
> > users can no longer write to the drive, but root can still log in and fix
> 
> It also reduces file fragmentation, so you dont need to 'defrag'.
> The reserve should be 10%. Remember that a 17GB HDD is really only
> 15.8 GB, the rest is added by Marketing.

When I was at school, in the 50s & 50s, kilo meant 1000 (as in kilometre), 
mega meant 1000000 (as in megahertz).

After some 30+ years' computing, I have to concede the computer industry's 
redefinition isn't entirely rational.

I did see one argument that
k means 1000 (a little like short tons)
K means 1024 (a little like tons or long tons)

For those lately arrived on Planet Earth (at least, this part of it)
	 a ton is an archaic measure of mass & weight. It's 2240 pounds weight, a 
short ton was 2000 pounds weight. A ton is close enough for most purposes 
to a tonne.

Back to disks. Years ago, disks were often soft-sectored. One could format 
them to one of several sector sizes: powers of 2 from 128 to 4096 were 
commonly available.

An enormous advantage of larger sizes was that one reduced the space lost 
to inter-block gaps; with 128 byte sectors, wasted space approached half 
the theoretical capacity of the disk.

When I was using OS/2, I sometimes wondered what would happen if I 
reformatted my drives with, say, 4096-byte sectors which experience (on 
mainframes) suggests should make good use of the possible disk capacity 
and perform pretty well with sequential reads (such as when loading 
enormous programs and some of the enormous documents certain 
wordprocessors create).

Note that picking optimal block sizes can be a little tricky. I used to 
use disk drives with 13030 bytes/track, others 19090 (or thereabouts). 
While blocks of a little below 13030 make good use of the first drive, the 
same size wastes about 30% of the second.

Anyway, when I bought an MO drive with 2k sectors, I discovered what would 
happen. The drive would have been unusable, at least with HPFS. 
Fortunately, IBM was working on the problem at the time, so it didn't 
remain a problem for long.


Anyway, I don't think we can rightly say the marketting folk are wrong to 
equate k with 1 000, m with 1 000 000 and g with 1 000 000 000, t with 1 
000 000 000 000 - that convention long predates computers.

-- 
Cheers
John Summerfield
http://os2.ami.com.au/os2/ for OS/2 support.
Configuration, networking, combined IBM ftpsites index.