About Gilmore's letter on IBM&Intel push copy protection into ordinary disk drives

[Bill Stewart]

At 04:10 PM 12/24/00 -0500, Steven M. Bellovin wrote:
>"Standard"?  It was more than that; it was the *right* thing to do.  On 
>a diskless workstation, there was no other identity to the machine; if 
>you didn't swap the ID prom -- which was used for the low-order 24 bits 
>of the Ethernet address -- your machine wouldn't receive the proper 
>boot image, etc.  Add to that the number of machines in the mid-to-late 
>80's that didn't have ARP, and it was utterly necessary.

But it was only the right thing to do if your the spare machine
was roughly identical to the dead machine.  Fortunately,
Sun operating systems are pretty good at guessing hardware,
due to S-Bus self-identification and the relative consistency of 
hardware.  With diskful machines, you sometimes had to worry about
whether the spare machine had as much disk as the original, 
though diskless machines didn't have that problem.
Sometimes you'd get into trouble because the spare machine
had a wimpier graphics board than the dead machine,
and the graphics configs might be stored in the copy of the program
that was on the file server, but usually you were ok.

You could also get into trouble if Box A (now deceased) had Program A,
and Box B had Program B, so you couldn't move the prom from A into B
without making it unable to run Program B.
But of course, if you've got a $20,000 chip-design program,
and $5000 Sun workstations, it's more cost-effective to keep
spare workstations around than spare program licenses,
and _much_ more cost-effective to keep spare disk drives.



				Thanks! 
					Bill
Bill Stewart, bill.stewart at pobox.com
PGP Fingerprint D454 E202 CBC8 40BF  3C85 B884 0ABE 4639