[rescue] Perverse Question

[Sheldon T. Hall]

Bill McDermith writes, in response to my tale of master-account-record
editing in days of yore...

> My first "real" (as opposed to summer jobs in school, etc) was
> working for a little local company that was contracting for
> Citibank, NA. It started out as a letter of credit system that
> was later expanded to do demand deposit, etc. All done on PDP-11/70s
> using Tek terminals. Whole system written in Basic Plus and
> Basic Plus 2 (you need bp2 to access the RMS record management
> library) on RSTS/E. All sorts of redundancy built in to reduce
> the number of transactions that would be missing if things
> crashed -- identical transaction logs identically written
> to several disks (sort of an early RAID concept, assuming the
> I means Identical, not Inexpensive :-)

The state of the computer art had progressed a lot between the NCR 315 and
the PDP-11.  We had neither disks nor tape, unless you count the paper tape
reader.  The PT reader had vacuum tubes in it.  The memory, all 56k of
12-bit "slabs", was real cores wound on little bobbins, 8k of them to a
cabinet.  All the connections were, as far as I could tell, hand-wired.

There was a _lot_ of mechanical pocketa-pocketa involved.  So much so that
we had a live-in NCR technician to keep it going.  The CRAM units were
amazing in their Rube-Goldberg-ness; there's a reason why that technology
isn't found in computers after disks got to be even half-way reliable.

FWIW, if you ever go by Stanford U, in Palo Alto, California, stop in the
Gates Computer Science building and look at the platter out of Stanford's
first disk drive.  About 4' (four feet) in diameter, an inch thick, and
solid steel.  The outer third of it is all chewed up from a head crash.  It
stored 4 MB per platter side, I think.  The attendant blurb about the disk
drive said it took 2 hours to spin up and stabilize, and I expect that
gyroscopic precession was a problem.

We've come a long way, baby.

-Shel