app requires C: less than 1.9GB?

Wed Mar 19 01:07:29 PST 2003

N. Thomas wrote:

>Someone just asked us puzzling question. They have a medical application
>that runs on Windows (don't know which version), and requires that the
>size of the C: drive it is installed on to be less than 1.9GB.
>
>We've never heard of a size constraint like this and are puzzled as to
>why a maximum size was specified (as opposed to "you must have N bytes
>free"). The nearest that we can come up with is this: 1.9GB is just shy
>of 2^31 bits, so they must have something hardcoded into their
>application that can't address outside this range.
>
>Anybody got a better explanation? My friends and I were just curious
>about this, that's all.
>
>thanks,
>thomas
>
>  
>
Having been a Product Manager for an industrial PC mfgr, I can tell you 
that there are lots of embedded applications that only need a 300 MBy 
HDD for the entire OS and application..  When a big OEM goes into 
production for a machine that they will be shipping lots of boxes of, 
they might buy several thousand drives.  The life cycle of a HDD is 
typically no more than 6 mos and most mfgrs won't bother you with an EOL 
notice unless you have a huge run rate.  (LCD screens are about the 
worst - you place your order before it goes into production and by the 
time you get the new product you can no longer order any more of those 
units and you are already getting eval units and looking at its 
replacement.  Life's a real bitch if you didn't guage your production 
accurately.)  

A 'tried-and-true' model can be very important.  There can be unforseen, 
ugly surprises by changing model numebrs.  We got burned once by a drive 
HP (I think it was HP - before they got out of the drive business) made 
that the head would somehow 'weld' itself to the media after use and 
shutdown.  This would prevent it from spinning up and if it did spin up 
it was toast anyway.  For whatever reason it took IBM a while to 'remedy 
that situation.  I had several customers who had married to that drive 
and were all over my pink butt about it.

Anyway, big OEMs often don't give a rat's ass that a drive 10X the size 
for half the price is available.  They want to ship the same drive for 
ever and ever amen.  If they have gotten any 'approvals' then changing 
an element like a drive can mean a whole new round of 
testing/certification and that can be an easy $10,000+.  FDA is one of 
the absolute worst to have to go through approval ffor.  NEBS is another 
bad one.  If you are going with full NEBS you pipe propane into the unit 
and burn it up to see what happens.

They also want to be assured that they know EXACTLY what the technician 
will find under the lid when he looks inside.  If he/she travels half 
way around the globe and is sent with a drive that for some odd, 
unforseen reason won't work in the box, a customer can get real pissed. 
 Some OEMs need very tight control of what components they use and spare.

So I don't know what this customer's situation is but they might have a 
couple thousand small drives sitting around they need to consume.  Maybe 
there is a flash device involved you need to stay within.  Maybe they 
run everything out of a RAM disk to try and control response times. 
 There are all kinds of reasons for it.

It would be interesting to hear what the reasons are.  They probably 
have real good reasons for it but it doesn't hurt to drill into it a 
bit.  Sometimes the customer is unconsciously married to certain 
concepts/technologies and you might bring some different approaches to 
the table.

good luck