WORLD CONFERENCE ON RECORDS AND GENEALOGICAL SEMINAR STORAGE DEVICES--THEIR POTENTIAL AND RELIABILITY

• Document Type: CREST
• Collection: General CIA Records
• Document Number (FOIA) /ESDN (CREST): CIA-RDP73-00402R000100140008-1
• Release Decision: RIFPUB
• Original Classification: K
• Document Page Count: 10
• Document Creation Date: December 19, 2016
• Document Release Date: July 19, 2006
• Sequence Number: 8
• Publication Date: August 5, 1969
• Content Type: REPORT

Approved For Release 2006/09/25 : CIA-RDP73-00402R000100140008-1
WORLD CONFERENCE
ON RECORDS
AND GENEALOGICAL SEMINAR
Salt Lake City, Utah, U.S.A.
5-8 August 1969
STORAGE DEVICES - THEIR POTENTIAL AND RELIABILITY
Part II
Digi+al Photo Devices
By GarrefBoer, B. A.
"Record Protection in
an Uncertain World"
COPYRIGHT 19a~j~ ~?~1 {~Qleag9 29 P9/25 : CIA-RDP73-00402R000100140008-1. r
A ?I S 1ETY- THE CHURCH OF JESUS CHRIST OF LATTER-DAY SAINTS, INC. NKI?-A ki - 8 & 9 b
 Approved For Release 2006/09/ ? CI--fPE7 ,:~(~J0?AR ~1~~11A40?ph- ,
STORAGE DEVICES - Hi
BACKGROUND AND HISTORY
Part I I
Digital Photo Devices
By Garr* Boer, B. A.
The Lawrence Radiation Laboratory in Livermore, California, is
operated by the University of California under the auspices of the Atomic
Energy Commission. It is devoted chiefly to the development of atomic
weapons and secondarily to a variety of other research projects most of which
require the use of high speed computers. To support this activity, the Lab-
oratory maintains a very large computing facility.
Several years ago, in the hope of improving the service of the computing
center, a contract was let to IBM to build a machine which could store a
great deal of information and make it available in a very short time. The
result of IBM's effort was the IBM-130`0 photo-digital storage device. This
device can store more than one-hundred pillion characters and can make
portions of this data available in a few seconds. We hope that its use will
materially reduce the vast amount of tape storage and tape handling which
is now necessary to keep the computers properly supplied with data.
METHOD OF RECORDING.
Notice. that the IBM-1360 is a photo-digital storage device. It does not
employ a photo-reduction process such as microfilming. Instead, it records
basically digital or numerical information by exposing film to an electron beam
which moves under the control of data supplied by an associated computer. It
is not, of course, difficult to convert alphabetic information to some numeric
code, so that records and documents may also be recorded and stored by the
1360. The 1360 must, however, be used in association with a computer which
supplies it with the necessary commands to control its operation. In addition,
the computer must supply the data which is to be recorded and collect from
the 1360 the data which is to be retrieved.
1
Approved For Release 2006/09/25 : CIA-RDP73-00402R000100140008-1
 Approved For Release 2006/09/25 : CIA-RDP73-00402R000100140008-1
VOLUME OF STORAGE
The basic element on which data is recorded is a small piece of film
called a chip - about 35 by 70 millimeters or roughly twice the size of a
35 millimeter slide. One chip can hold nearly five million bits of data or
nearly six hundred thousand alphabetic characters if the currently popular
encoding of 8 bits per character is used. This is enough encoded alphabetic
information to record more than 500 pages of the New Yorker Magazine.
For storage in the 1360, thirty-two of these film chips are placed in
individual slots in a little box called a cell. A cell is about one inch by one-
and-a-half inch by three inches, so that you may comfortably grasp in one
hand about thirty-two volumes of 500 pages each. The 1360 configuration
at Livermore can store nearly 7 1 000 of these cells for rapid access by our
computers.
MECHANICAL DESCRIPTION
The 1360 is a large and complex collection of mechanisms which has
facilities for automatically exposing, automatically developing, automatically
storing and retrieving film chips and finally for automatically recovering
and checking data from these chips. A variety of configurations is possible.
The one at Livermore occupies about four hundred square feet and has a
station for recording and developing, two stations for reading or recovering
data, three storage files to hold the cells and a control section to coordinate
the device's myriad activities.
The chips and cells are manipulated and transported in a manner which
suggests that Rube Goldberg's drawings provided the inspiration. The unit
of transportation is the cell, which serves as a bus to carry its thirty-two
chips from one place to another; for example, from a storage file to a reading
Approved For Release 2006/09/25 : CIA-RDP73-00402R000100140008-1
 Approved For Release 2006/09/25 : CIA-RDP73-00402R000100140008-1
station. This bus is propelled through airtight tubes by differences in air
pressure. The transit system uses the same principle of operation that has
been employed by department stores to send your money and bill to a central
station and to return your change and receipt.
When a cell is to be removed from a storage file, an elaborate mechanical
arrangement of trays and tubes is shifted about until one of the 2250 cells in
that file is lined up in a pneumatic tube; a burst of air then propels it to its
destination, which may be ten or twenty feet away.
When a cell arrives at a reading station - a few seconds later - another
sequence of elaborate mechanical actions slide off the lid, remove the desired
film chip from its slot and position the chip in the path of an electron beam
which will scan it and recover the recorded information.
A chip being recorded is likewise subjected to a complicated sequence
of operations. An unexposed film chip is removed from a cell containing raw
film and placed in a chamber where it is subjected to a partial vacuum to de-
gas the surface. Then as the film is moved into the recording chamber,
another chip of raw film follows into the de-gassing section. In the recording
chamber the film is exposed to the moving electron beam to record the desired
information. After exposure this chip moves out into the developer and the
following chip is moved into the recording chamber. Thus, a continuous
queue of chips is kept stepping through the recorder.
This queue of chips then moves through a series of stations where they
are developed, fixed, washed and dried. They then emerge one at a time and
are placed in a cell for storage.
The time required to record and process a chip is about three minutes,
but many chips can be in different stages at the same time, so that a finished
chip maybe produced about every eighteen seconds.
3
Approved For Release 2006/09/25 : CIA-RDP73-00402R000100140008-1
 Approved For Release 2006/09/25 : CIA-RDP73-00402R000100140008-1
The 1360 is designed to allow many of its various activities to proceed
simultaneously. At any one time several cells may be queued up at each
of our two reading stations, waiting to deliver their chips, each of the three
storage files may be rearranging itself to locate a particular cell position,
a cell may be in transit.from some point to another and, of course, several
cells may be underway in the recording-developing queue.
Control of all this activity is accomplished by the use of a special
computer which is an integral part of the 1360. This process control computer
must not be confused with the external computer which is still necessary to
direct the overall activity. The internal computer interprets commands
from the external one. It then sets in motion and keeps track of the dozens
of electrically controlled mechanical motions which must be performed in
order to obey each command.
1360 ERROR CHECKING AND RELIABILITY
There is, of course, no point in all this activity if the recovered data
is not reliable. There are several aspects to the problem of reliability -
one is the 1360's internal checks on the data. The 1360 is designed to generate
a good deal of redundant information dependant on the pattern of the externally
supplied data and to record this redundant information alongside the supplied
data. Then, when a chip is retrieved and read, this redundant information
is regenerated and compared with that which was previously recorded.
This comparison allows the 1360 to determine with a high degree of prob-
ability whether the recorded data is correct and even allows the reconstruction
of correct data from bad data if the errors are not too severe.
Rather than wait until someone wants to retrieve the data before
learning whether or not the 1360 can successfully read a chip, it is good
practice to attempt a check-read soon after the chip is recorded - before the
4
Approved For Release 2006/09/25 : CIA-RDP73-00402R000100140008-1
 Approved For Release 2006/09/25 : CIA-RDP73-00402R000100140008-1
external source of the data is discarded. Then it will be possible to immediately
re-record a bad chip.
The errors which can be detected internally by the 1360 seem to arise
chiefly from small imperfections on the surface of-the film which are produced
during its development. Our experience at the laboratory so far seems to
indicate that we can expect that an average of about one chip per cell - or
one in 32 - will have to be re-recorded.
The error-checking described above is strictly an internal check. If
incorrect data arrives undetected at the 1360, the redundancy information
will be generated from that bad data and when the check is made later, the
1360 will give no indication of error. To further check the data it would
be necessary to use the external computer to recover the data from the 1360
soon after recording it and to check it -bit by-bit against its source.
Once the chip is reliably recorded, it is presumably good forever. We
do not yet know whether the chip will indeed deteriorate significantly during
a long period of storage.
THE LABORATORY DATA NETWORK
The Lawrence Radiation Laboratory maintains a large computer center.
The bulk of the computation is performed currently by a group of powerful
Control Data Corporation computers: four CDC 6600's and one CDC 7600.
An extensive data network has been set up at the laboratory for at least
two purposes.
The first purpose is to connect these computers to inquiry stations.
At present there are a couple of hundred of these stations and each one con-
sists of a teletypewriter. We expect the number of stations to grow and we
also expect to have more elaborate equipment at each station, for example,
a cathode ray tube or TV-like display. From any teletypewriter in the system,
5
Approved For Release 2006/09/25 : CIA-RDP73-00402R000100140008-1
 Approved For Release 2006/09/25 : CIA-RDP73-00402R000100140008-1
a user may communicate with any computer in the system and may initiate
some work on that computer.
The second purpose of the data network is to connect each of the basic
work-horse computers to the IBM 1360 and to another storage device, the
IBM Data Cell, so that data originating at a computer may be stored on one
of these devices and the stored data may be retrieved by any of the computers.
At the center of this network is another computer - a Digital Equipment
Corporation PDP-6. This computer has data channel connections to each of
the larger work-horse computers, to each of the storage devices, and in-
directly (via two smaller computers) to each of the teletypes. In addition,
the PDP-6 is attached to a high speed disk. This disk is an intermediate
storage device which is used to hold data temporarily while it is en route
from one place to another.
Before the advent of the IBM 1360 and its incorporation into the network,
a computer user had two ways to make large files of data available to his
program. First he could enter a deck of cards through the card reader.
Secondly, he could type in a message on the teletypewriter. This message
would be relayed to a human operator who would in due time seek out and fetch
a particular tape reel from our voluminous vaults and hang this tape on a tape
unit which is connected to the appropriate computer.
Today, a user may type a message on his teletypewriter which is
passed on to the central PDP-6 computer. This computer issues commands
to the IBM 1360 and begins the transfer of data from there to the attached
high speed disk. When this transfer is complete, the PDP-6 gains the attention
of the appropriate peripheral computer, and with its cooperation begins trans-
ferring the requested data from the disk to that computer.
Approved For Release 2006/09/25 : CIA-RDP73-00402R000100140008-1
 Approved For Release 2006/09/25 : CIA-RDP73-00402R000100140008-1
The actual data paths are somewhat more devious and a good deal longer
than this description indicates. In fact, a teletypewriter message or a data
file can follow a route traversing a dozen different electronic devices made
by half a dozen different manufacturers.
NETWORK ERROR CHECKING AND RELIABILITY
The lengthy data routes indicated above create additional reliability
problems outside of the storage device itself. The reliability checks pro-
vided for in the system hardware tend to be very local checks. There is
nothing comparable to the elaborate redundancy data incorporated internally
in the 1360 which can be used to check the data over the entire transmission
path. Some less elaborate redundancy checks are provided at various places
along the routes and these do catch many electronic failures and provide a
warning to try again. However, we know that many data errors do go un-
detected by the hardware and that retrieved information is found to be
different from that which had been sent to storage.
Programming techniques can be used to a limited extent to overcome
this difficulty. Some relatively simple redundancy calculations can at least
provide a high probability that you will receive advance notice that some
error exists in the data. More elaborate calculations which. would allow re-
construction of erroneous sections would require too much computer time to
be feasible.
DOCUMENT STORAGE AND RETRIEVAL
The file storage system in use at the Lawrence Radiation Laboratory is
not specifically designed to meet the requirements of those who are interested
in storing and retrieving documents, and my own work has given me no ex-
perience in this application. There are, however, a few general observations
I can make about the types of equipment that would be helpful in exploiting a
large digital storage device for this purpose.
7
Approved For Release 2006/09/25 : CIA-RDP73-00402R000100140008-1
 Approved For Release 2006/09/25 : CIA-RDP73-00402R000100140008-1
First, of course, a sufficiently powerful computer must be employed
to control the storage device and to handle the flow of information in both
directions. There should also be connected., to the computer an auxiliary
storage device such as a disk or a drum. This auxiliary device will have a
good deal less capacity for storing data than the 1360, but could make its
data available much more rapidly. It will be useful for at least three purposes:
First, it can hold part or all of the index information which is necessary to
locate documents on the larger storage device; secondly, it can temporarily
hold data which has been retrieved from the 1360, so that documents may be
printed from this source while additional retrievals are in progress; and
thirdly, it can hold data which has just been recorded until the read-back
check'is complete, so that the data is available for re-recording if the chip
is bad.
One essential part of such a system will, of course, be a card reader,
so that documents may be punched on IBM cards for entry into the computer.
There are several considerations which govern the selection of equipment
to aid a user of the system in searching for and retrieving a document from
some inquiry station. First, it may be desirable to have several inquiry
stations so that many people may be simultaneously occupied in searching. A
keyboard of some type, such as that on a teletypewriter, is necessary for
entering requests into the computer. Also, some searches may require a
good deal of conversation with the computer; that is, some initial information
is returned by the computer and this prompts the requestor to ask for another
retrieval to bring him nearer to the ultimate object of his search. For appli-
cations like this, a teletypewriter may be adequate for both entering the request
and for printing the computer's response. If, however, large portions of material
must be scanned before making another request, a teletypewriter might seem
Approved For Release 2006/09/25 : CIA-RDP73-00402R000100140008-1
 Approved For Release 2006/09/25 : CIA-RDP73-00402R000100140008-1
frustratingly slow. For these applications, a device which can display infor-
mation more rapidly is desirable at each inquiry station. A Cathode Ray tube
or television display would adequately meet this need. The system user could
scan a few paragraphs and enter a new request via the keyboard.
The TV tube might not, however, be suitable for a more relaxed or
lengthy perusal of the retrieved document. Besides, a printed copy is usually
wanted for subsequent reference or for use somewhere outside the computer
room. To meet this need, a single high speed printer might be tied to the
computer to produce this hard copy for all the users of the system.
The advent of the IBM 1360 has moved us into the era when massive
collections of data can be speedily accessed. This report is meant to provide
a little insight into its workings and some of its problems. I have only hinted
at the extent of the programming effort necessary to exploit such a device.
It should provide many interesting challenges to those who wish to make use
of it.
This report was prepared as an account of Government sponsored work.
Naler the United Stales, not the Commission, nor any person acting on behalf
d the Commission:
A. Makes arty warranty or representation- expressed or implied, with
respect to the accuracy, completeness, Of usefulness of the information coo-
Hind is this report, or that the use of any information, apparatus. method, as
process disclosed in Ibis report may vol infringe privately owned rights; or
9. Assumes any liabilities with respect to the am of. or for damages
resulting from the are of my information, apparatus, method or process dis-
closed in this repot.
As ssd a the above, "person acting on behalf of the Commission"
helvdes any employee or contractor of the Commission, or e~pleyee or such
Couhac a, to eta extent that sacs employee or contractor of Purr Commiusrah,
a employee of sack contractor prepares. dissewloates, or provides access as.
/4 idoemati . porsuaol Of his employment at contract with me Commission,
a his employmed with such contractor.
9
Approved For Release 2006/09/25 : CIA-RDP73-00402R000100140008-1