The US Veterans Administration Data Processing Center in Austin, Texas had
*at least* three -- maybe at one point five -- of these beasts (one on a
360/40, two on a 360/65, maybe two more at one point on a 360/50)..
I was an operator on those systems at the time -- late 1960's.
Although speed was a big problem with the (as I recall) 2321 Data Cell
Drive, a bigger problem was reliability. They had a nasty tendency to
"crunch" strips, and when that happened, the strip had to be recreated
from
backup. This was virtually a daily occurrence in our shop. I think we
had
a fully-2321-trained IBM engineers on site for two of the three ****fts,
and
another on call for the third ****ft, seven days a week.
The first heavy production jobs -- I can't remember whether it was
logistics
& supply or VA ****tfolio loans -- took something like ten hours for the
daily processing on a 360/65, if it ran successfully. It didn't usually
do
so, and after a month of trying we were a month behind. At that point,
each
400mb 2321 was replaced by a pair of "pizza-oven" 2314's with about the
same
400MB capacity, and much greater speed and reliability. Daily processing
dropped to three hours or less, and restorations from backup were almost
unheard of.
Part of the difficulty, I think, was that they used the devices for ISAM
files with the indices in three of the ten bins and the data in the other
seven, so every record access required a "double pick", one on each side
of
the array of strips. This was an application design issue; the device
works
fine for "online inquiry", but not as a batch DASD device.
I think the Texas Department of Public Safety also had one or more of
these;
from what I heard at the time their (occasional online inquiry)
application
was better suited to the device than the VA's were.
-Chuck Stevens
<mauzbiz@[EMAIL PROTECTED]
> wrote in message
news:1146781586.864406.79400@[EMAIL PROTECTED]
> Worth remembering for "musem-quality" items is the mid-1960's IBM
> random access "data cell", scarcely known and little deployed, that was
> like a round carousel juke box that spun hundreds of magnetic strips
> around until the chosen one was wrapped around a read/write head from
> whence data could be read or to which data could be written. The
> device had a capacity of 400 megabytes, far more random access storage
> than most shops had online at any given time, and could access any data
> in a few seconds. But it was expensive and I suspect attracted little
> interest because most applications in those days were happily running
> with an input tape, transaction tape or cards, and an output tape, and
> companies were only beginning to explore random access files for
> certain limited purposes, given the cost per megabyte.
>
> RCA had a version of this type of device, which, under certain
> cir***stances of motion (back and forth), would inadvertently drop some
> of its strips out the back of the machine, and some clever techie
> programmed such a sequence and then announced to the operator "I've
> lost one of my strips, could you please come over and help me out."
>
> I remember our IBM data cell from a summer job at California Western
> States Life Insurance Company in Sacramento in 1966.
>
> It would be fascinating to bring into a musem such a first attempt at
> leapfrogging the extremely limited 7 megabyte limitation that most
> random access devices (usually removable disk pack drives) had even
> into the late 1960's. Like bubble memory, its capabilities were
> surpassed by other technological improvements and it never found a
> significant niche.
>
|
