(Sanitized) BRIEFING ON THE CYPRESS CHIP STORAGE AND RETRIEVAL SYSTEM, 10 AND 11 JUNE 1965 AT SAN JOSE', CALIFORNIA

• Document Type: CREST
• Collection: NGA Records (Formerly NIMA)
• Document Number (FOIA) /ESDN (CREST): CIA-RDP78B04770A002900030024-3
• Release Decision: RIPPUB
• Original Classification: C
• Document Page Count: 25
• Document Creation Date: December 28, 2016
• Document Release Date: August 26, 2005
• Sequence Number: 24
• Publication Date: June 18, 1965
• Content Type: MFR

? Approved For Release 2005111121 :CIA-1 V7$~ptlf1UA002900030024
~'N~L
3. I attended on invitation ofi of OCR. Their visit was
discuss a solicited proposal from to provide the video CYPRESS sys-
as the Agency mass document storage and retrieval system from hard
o JJ" in--- to hard copy output. Indexing, however, was not proposed. It
was ass-.:-.~_l that, by other means, the address of the chip of a wanted
focumen-;; _ _, k;.nown. Those attending the meeting were:
18 June 1565
Declass Review by NGA.
Briefing on the CYPRESS Chi.--, St-ora--e and ',Retrieval
System, 10 and 11 June 1965 Jose `, California
1. CYPRESS system began as a photo chip m_;:.
s system wf th ~erat-~ 'e ca_ as ns o.. It
nog, ir_c _na _ ; odigatal record::- and video recording al tbo uL ,;~ _e
la,.Ler :vastly in the conceptual stage.
2. Continent: Image CYPRESS is costly and may be slow. Careful
S-3u-ay of a specific use should be made before purchase. Simpler methods
will of-cen suffice. Maintenance will prove costly just as in the Mini-
ca d system (some per year for a small system).
P'_-_`..w-gital CYPRESS will probably be packed for large systems (not
c,.__ed for small systems) which reduces the importance of the chip
u_n .: cord. The Osystem uses a much larger disk record with
1a.., try-train-switching and plurihling. Random access is an
Bali.." siren, sitting on the rocks of wasted space (per chip) and
.._eauad average access time; it charms one away from the problems of
organization only to throw one rapidly back into the problems in
o_der to manage batch retrieval efficiently. One-at-a-time retrieval,
although needed, is seldom efficient. Very few cells can be in movement
at once.
Video CYPRESS is impressive as a concept, but the chip mechanics
main the least impressive part. As a reality, it remains to be seen.
Its progress should be carefully monitored. I can not recommend more on
the basis of this exposure.
To repeat, the chip decreases as a logical-sized unit record as we
go from image CYPRESS, to photodigital, to video.
Enludad Pram automatic
Approved For Releases~ r ,, C IAFR,P 70A002900030024-3
Ln
25
25
 rRrrd LU06E'ONLY
Approved For Release 20 1~1 Td -K DP78B04770A002900030024-3
X1
X1
X1
SUBJECT: I iBriefing on the CYPRESS Chi:D Storage and Retrieval
System, 10 and 11 dune 1965 at San dose', California
California,
4. I I of SSD reviewed the photo image CYPRESS system
and the photodigital system. The photo image system is quite similar to
the plans as I reported them two years ago (8 May 1963). They now plan
8 images per chip (a 240 reduction only. Input will still be via micro-
film so they can use Diazo film as the storage chip. Output is still to
Mil D Diazo aperature cards by contact printing. Hard copy can be made
from the aperature cards.
32 chips per cell = 256 images
450 cells per tray = 115,200 images
5 trans per file = 576,000 images
3 to 5 files in a system
X1
Twenty-five feet per second cell delivery speed in the pneumatic sys-
tem was quoted, or 1000 cells per hour. A single tube delivery system is
planned, with switching and delays to return cells to their home location
and this will probably control the effective size of the system. They
pointed out that parallel tubes were possible, but did not indicate they
planned to go this route.
This is still referred to as the engineering drawing system. It was
found to be the most costly of some 20 systems examined by the CHIVE group.
5. The Photodigital CYPRESS is new. However, two systems
They will be used for mass digital data storage to replace magnetic tapes.
Digital bits will be recorded on the chips photo optically - somewhat a'la
I have been sold to the AEC for delivery in about 2 years.
Minicard but more like the photoscopic disk work being done by
They still plan:
25
25
25
25
It employs electron beam recording on special
DER film made by Eastman (Digital Electron Beam Recording). It has a 7.5
mil ESTAR base with a 5 micron emulsion, very hard. It has a high silver-
to-gel ratio; no optical sensors, thus "insensitive" to room lighting; is
somewhat similar to EK 61-9GH film. The recording beam is 2 micron wide.
2
Approved For Release 200 /81 / - ill I ,~ )174$14f jKpA?02900030024-3
 Approved For Release 2/111:AtR4WE~9A!LYA002900030024-3
SUBJECT: I riefing on .. e Cr ?ASS Chip Storage and Retrieval
Syste`, 1C and :-y J uane 1965 at Sun Jose T , California
Because the recorded data cannot be read until the chip is
developed, the read-back method cC accuracy-checking used on magnetic
media is not available. High redundancy recording is used instead. Every
data bit is recorded twice, plus some 66 check bits per 300 data bits.
These are used with elaborate error correction and detection techniques
at readout time. 4,000 core positions are required for these techniques.
They can detect and correct up to 5 six-bit charact,.aw?s (can detect a sixth,
but cannot correct it nner line on a document. But note such detection
and correction must be done every time the chip is read out. The techni-
ques include matrix inversion.
5,C30,000 bits can be recorded on one chip. At 6 bits per character,
this iu 833,333 characters. However, for digital recording, the bits per
character is up to the user; it's basically just a string of bits. One
file. holds 1/3 trillion bits. The AEC configuration will hold 2.5 trillion
bit--, (3 file nodules)*, They will have 1 controller (electronic control
cor.:-)uter), 2 input stations, and 1 output station.
1.5 second cell-retrieval time was stated or 3 second retrieve and
return. They plan to overlap the return.
Readout is 2 megabits per second.
Input can be temporarily recorded on a disk file to be immediately
available for retrieval and to be later replaced with the developed chip.
They record at 500 kilobits per second or 21.5 seconds to fill an entire
11
chip (5,000,000 data bits)X*.
They hope to be able to contact print duplicate photo digital chips
.ter, permitting reproduction of data stores.
6. The Video S stc-:m was presented by several ASDD engineers and a
sum-up was made by of SDD. Several items described by
ASDD are not yet offered for delivery.
explained the input method.
* File Module: 5 trays in first one, 10 trays in succeeding ones.
T ~TYis seems to bear out my previous figures:
for 300 data bits, each is recorded twice plus 66 bits for error
checking = 666 bits or 2.22 bits per data bit, 5,000,000 data
bits per chip = 11,100,000 total bits. At 500,000 bits per
second record speed, a full chip would take 22.2 seconds.
Approved For Release 2005/11/21 :3CIA-RDP78BO477OA002900030024-3
ONLY
 Approved For Release 2005/11/21: CIA-RDP78BO477OA002900030024-3
CIA INTERNAL USE ONLY
SUBJECT : I L^iefi ng on the Cam' = ESS C _ `.'-orage and Retrieval
Syste ri, 10 and 11 June 1965 at San Jose', California
A suecial scanning disk will scan micro?itm images of documents (sil-
ver haloid, diazo, or -Kalvar) and digitize the data. The data rate is a
function of the scan disk speed. Expect about 4 seconds per chip.
They are also experimen-cin_g with a line scanner to be used on hard
copy. they have experimented with 100, 150 and 170 lines per inch scan-
ning. Brace compression technique is used as .follows:
When data (mixed black and white spaces in close '_croximity) are scan-
ned, thay are recorded as 6-bit data records, the first bit being reserved
to indicate that the word is a data word. 117hen clan! space is encountered
(CC:_ti ui g black or continuing ?te), the first bit indicates this is a
s Dace-cor.. ession .?rord, the second bit indicates whether white or black
space is encountered, and the remaining 4+ bits record the number of data
biucs being compressed. This variable length of space compression even
picks u-r oases between words. At 100 lines per inch, compression ratio
about 3:1, at 150, 5.5 to 1.
Abo Lt 126 characters are recorded to identify a single alphameric
typed c :aracter
This plan is black and white only; no grey scale.
Output would reverse the process, scanning the chip and depositing
electrostatic charges on paper for Xerographic development. They scan at
2700 characters per second (I guess they mean the 6-bit word; this is not
clear to me) or 2025 lines per minute; 2 seconds for an 82 x 11 page.
This output description requires an unscrambler which I am unable to
supply. went on to describe a 14-8-character set and a capability
of expanding to lower case characters. I suddenly found him describing a
character recognition recording and print-back technique. Later he demon-
strated print-back only. I believe that this character printing was only
a laboratory technique to establish certain criteria for their further
work. It was not described as part of the ultimate system. I was sur-
prised by the implication that we had received a demonstration of the
technology involved".
. Such scanning can, of course, also be delivered to a TV viewer. This
was demonstrated by loops of magnetic tape being read (iteratively about
30 times per second) and "printed" on a TV screen. Pictures were shown
too. The demonstrated TV technology is well within the current state of
t e art. Input to chips was unconvincingly demonstrated and there was no
throughput demonstration; that is, the output scanning of a chip and
recording it on the tape loop was not shown.
Approved For Release 2005/11/21 ; CIA-RDP78BO477OA002900030024-3
CIA IAN I LE N'A L Ut;E ONLY
 Approved For Relaa%e 1A9~-~R
~~ lL .9~18i0477OA002900030024-3
X1
X1
S S ECT. I-Kis c on C_. '":JS Chip #torage and Retievn._
-
Sys uen, 10 and 11 June 1965 at San Jose ' , California
file and display devices. We were told that at about 2000 scan lines per
page, both input and output was possible at a rate of 10 to 30 paves per
second.
7. syske further on the video
They have made a 3000 line flying spot scanner. A 940-line 20 mega-
cycle scanned image of a page was shown; it was of poor quality,
At 10 pages per second input and output; they will use a buffer to
_. egenerate _rages for TV viewing (?he _iagneti~ to e loop is one type of
buffer). One buffer can drive several viewers. Or, a page of hard copy
can be scanner-"exposed" in 110 second.
A chip holds about 2 square inches of recording area on each surface
(front and. back). They -propose to use both. Magnetic chips can rub with-
out damage, thus 64 chips can be put in a cell, "probably more". Later,
however, we were informed that the proposal to CIA was based on 32 chips
per cell.
10,000 bits can be recorded on one 2" length of track on a chip, 100
tracks per side. Use a high density and high velocity 6-mil scanner;
600-800 ips. Four megabit transfer rate.
Can write one track, read, and check it in 16 2/3 milliseconds. The
limiting factor is moving from track to track: 7 milliseconds. Read and
write 600,000 bits per second; read, only, at twice that rate.
Reading and writing at the 5000 bits per inch per track requires
physical contact of the read-write heads. They find that it takes 1,400,000
ra,oes to "7 gar out a track". "One track in 40 will have an error.''
It is possible to get grey scale, maybe 8-10 shades, possibly 16.
Bt only digital black and white recording is proposed to CIA. Grey scale
will require analog recording. "Color is possible."
The reading and writing takes place over a drum (see attached sketch
B). Note that, as shown in the sketch, the right-hand surface can be read
(or written onto) by opening the cell, picking the correct chip, and
pulling it over the drum. The process is reversed to return the chip to
its cell. To read the other surface of the chip, the cell must be moved
to position two (see red arrow) and the process repeated, this time
pulling the chip up counterclockwise over the drum. Because it appears to
be more rapid to read several chips on the same side than to read both
sides of one chip, then both sides of the next, etc., I asked if this were
the recording plan: record all chips on one surface, then all on the other
surface; or use this method per cell. I was confused by the reply but
inferred that they did not agree with my conclusion that reading first one
side, Approved ~or e~ease NIX cf SAY ~ A R Pj#~6A 60 11660 naf 6
 Approved For ReIq cJAFF9181304770A002900030024-3
X1
8. On Friday morning we received a summary byl and
others. At this time the attached sketches (A through G) were received,
with a caution concerning their proprietary nature. Their disclosure
outside CIA is not authorized.
Brie in: o_. tho :; r ASS : __ir Storage and Retrieval
System, 10 ant L June 195 at San Jose', California
Most of my notes about the sketches are recorded thereon.
Video printer: 40 to 50,000 6-bit charan,ars per second. About 10
seconds per page. Present plan is to display 1/3 of a page at a time;
buffer will hold 6 times this or two full pages. "it is possible" to
handle a full page at a time. May encounter a delay of 7 seconds to
access a new page. Can have 3 buffers and 5 printers per buffer, or total
of 15 printers which would use up all channels of the electronic system
controller (a small computer)
Currently planning 100 x 100 lines per inch. Demonstration of quality
at this density: very readable but I would expect eye strain to result
(obviously a personal reaction).
Printers and viewers can be up to 2,000 feet away from central unit.
Printer will yield 82" x 11" paper.
The core requirements for the controller are:
X1
image system 4)000 bits (nay grow)
photodigital system 8,000 bits
video system at least 16,000 bits
7 data channels
Proposed to CIA:
16,000 chip system, or about 1.2 million pages with average
compression of 3 times.
digital; no grey scale
9, CYPRESS Simulation: Overlap is logical and essential to get any
reasonable speed out of the CYPRESS system. The speed of the system, bot-
tlenecks, and ways to break the bottlenecks for different assumptions con-
cerning input volume and the volume and nature of output requests can be
simulated with a computer. 0 explained their use of the GPSS (I
assume General Purpose Simulation System) to study the problem. The simu-
lation results of several different mixes were displayed. The work is
most interesting and may be worth a closer examination. Recording the
results was impossible and nothing was supplied in writing, but there fol-
lows a sketchy description of the method.
25
Approved For Release 2005/11/21 :CIA-RDP78B04770A002900030024-3
CIA INTERNAL USE. 'ONLY
 Approved For Release 2005/11/21: CIA-RDP78B04770A002900030024-3
CIA INTERNAL USE ONLY
xi
Brief--'-.-,-,- c_, U ..?=SS Olin Storage and Retrieval
System, l0 and 11 u_e 1965 at San Jose Y, California
a. E1e ^e_,~s of System
Files
Video chip converter
Distribution buffer
Shuttle buffer
Document printer
Video buffer
Display
Scanner
b. Processinc Sequence
Scan Request
1/
2/ Queue for & seize Shuttle Buffer
3,/ Queue for & seize Cypress File
,f Position tray, transfer cell to
flip buffer
5/ Queue for & seize pneumatics
61 Transport cell to shuttle buffer
7/ Releast file & pneumatics
8/ Seize video converter
9/ Transfer cell to video converter
10/ Release shuttle buffer
11/ Pick chip
12/ Queue for Module Controller
13/ Use module converter to transfer
1 page
1L/
15/ Repeat steps 12 & 1-3
161 Queue for & Seize file
17/ Queue for & seize pneumatics
18/ Transport cell to file
19/ Release: video converter
pneumatics
file
Print Request
Queue for & seize Dist. Buffer
Same as Scan
Same as Scan
Same as Scan
Same as Scan
Same as Scan
Same as Scan
Same as Scan
Same as Scan
Same as Scan
Same as Scan
Same as Scan
Same as Scan
Transfer 3 pages
initiate printing
Same as Scan
Same as Scan
Same as Scan
Same as Scan
c. Job Set Up
Used random numbers for delay between requests (average
15 seconds). Ten seconds to print. 31 second delay on display
requests. 1 second average use time. Run time one hour.
Approved For Release 2005/11/21 :'CIA-RDP78B04770A002900030024-3
!1N * T" il i.:
U3 ONLY
 Approved For Release 2?O, /j'OJT' F8Ik:-Rl 40AO02900030024-3
X1
SUBJECT: riefing on the CYPRESS Chip Storage and Retrieval System,
L4and 11 June 1965 at San Jose', California
d< Data. Obtained
(Figures supplied for only a single simulation)
Element
No 0 of
Units
% Uti:Li-
zation
No.
Entries
Avg. Use _..Max..
Time Contents
Avr.
Contcnts
% Time in
Zeros Queue
Files
2
30
1150
1 or 230,704 chips or 1.6 million
legal-size images (@ 8 per chip). The system can hold up to 5 files. -(a
million chips).
J iZl kx- ~9T
Chips can be updated, purged, or re-arranged. They can increase
speed by proper chip filing and therefore plan to develop automatic chip
filing (or refiling) logic with an eye to rapid retrieval. Any "ordering"
of the file in lieu of random storage will probably mean dedicated storage
space which will reduce the storage capacity of the system.
OUTPUT: A search request goes to the 1440 computer which, like
WALNUT, searches its index and produces blank = Diazo aperture cards 25
containing chip addresses and other information (up to 57 columns). Az
"addressed" chip is retrieved by retrieving the applicable cell, opening it,
removing the chip, vacuum cleaning the image area (nice!), contact printing
onto the Diazo film in the aperture card, closing and refiling the cell,
and developing the print by ammonia under pressure. Developing takes 2
seconds and is overlapped with a 2 second exposure of the next chip. Out-
put speed: 1,000 cards per hour.
Input and output are combined at a single station and functions
are carefully overlapped for throughput speed. Requests and cells can be
queued by the computer. Up to 3 input/output stations (and 5 files) can
be handled by one system. They expect to have up to eight cells in
transit in the system at one time. Cells can be inserted or removed,at any.
input/output station.
VAPOR DEVELOPMENT: So far as I know, ammonia-vapor-pressure
development of Diazo film is a break-through (and this information must be
handled carefully as =proprietary). We saw a breadboard set up of a
piston for varying pressure, a cylinder of ammonia for metered application,
and a chip loading and holding device. I inferred that the 2 seconds
exposure time might be reduced, but that this equalled the time to remove
a chip from a cell and therefore met their objective.
VIEWER AND SMALL RETRIEVAL SYSTEM: They plan to build a viewer
with a 15" x 20" screen which will accept hand-loaded cells. It will present,
full size, two legal-size images at a time, will probably have other
magnifications, too. It will have a "joy stick" for moving the image
around on the screen.
There is some thinking under way for building a 100-cell input for
the viewer to serve as a small retrieval mechanism.
Approved For Release 2005/11/21: LUI X- 2 6B0477OA002900030024-3
 Approved For Release 2005/11 ,9IA-RDP78B04770A002900030024-3
3-
A small system would require a camera, punch, film processor,
and viewer. The index would be manually controlled. This is somewhat
would be used instead of small rolls in magazines.
0
systems men talked with:
C01MENTS: For high-quality photographs, I think it is a mistake
to have two stage input and output: hard copy to microfilm to Diazo chip to
Diazo aperture card to hard copy. For line drawings and textual material,
the degredation is probably not significant but the 2-stage input still
seems to be unnecessary. Using the cell as the movable unit instead of the
chip gives less flexibility than Minicard, more than WALNUT. You have to
open the cell to read a magnetic strip and here I believe Minicard will
be much faster. On the other hand, the Cypress and WALNUT ability to
address a single chip by specific location eliminates the Minicard need for
file expansion and permits efficient use of a computer for handling the index.
(I am at a loss to know why they want 100 characters on magnetic tape on
the chip when they can store information about the chip in the computer
without limitation.) yid
The idea of vacuum cleaning the chip image area prior to print-out
is excellent.
The small system with a viewer is interesting. I don't believe they
have done a lot of thinking about this, yet.
Inability to handle sets with ease appears to be a drawback. For
high quality photographs, 60X reductions are not acceptable.
If the Cypress chip were bigger Imight do a good job of building
a SCRAM storage and retrieval unit. Mundane things like costs and a
completion date were not mentioned.
)(- Imo. 1A1116 a a "
7L 12
25
25
Approved For Release 2005/11/21 : CIA-RDP78B04770A002900030024-3
 Approved For ReII YcY Ii I
la-e
-( ita ~r
a?e 115 Approved For Reel*?ej RO ft
:L -I B0477OA002900030024-3
 Approved For Release ?20O5i1l I2.t CAA-RDP78 ~7~7OA002900030024-3
 Approved For Release 2005/11/21,: CIA7RDP78B04770AOQ2900030024-3
AOL-
C'~-t ew.
Relj6se 2005/11}1 : CI RQ-70A002900030024-3
C(A INTERNAL (SSE ONLY
 Approved For Relea 220 ll~r
8g0477OA0029 0030024-3
f
ABCDEFGHIJKLMNOPQR STI;VWXYZOUN45 , . v , f
ABCDEFGHIJKLMNOPQR STUVWXYZQ123456789. .)+$*-/,(_'*
ABCDEFGHIJKLMNOPQR STUVWXYZ0123456789 .)+$*-/,(='$
ABCLEFGHIJKLMNOPQR STUVWXYZ0123456789 .)+S*-/,(_'$
ABCDEFGHIJKLMNOPQR STINWXYZ0123456789. .)+$*-/,(_'$
A13CE:EFGHIJKLMNOPQR STUVWXYZ0123456789; .)+$*-/,(_'$
ABCEEFGHIJKLMNOPQR STUVWXYZ0123456789 .)+$*-/,(_'$
ABCLEFGHIJKLMNOPQR STUVWXYZ0123456789 $
ABCDEFGHIJKLMNOPQR STOVWXYZ0123456789
ABCDEFGHIJKLMNOPQR STUVWXYZ0123456789 '.)+$*-/,(_'*
ABCDEFGHIJKLMNOPQR STI.IVWXYZ0123456789. ?. )+$*-/, (_' $
ABCDEFGHIJKLMNOPQR STUVWXYZ0123456789 .)+$*-/,(_'$
ABCDEFGHIJKLMNOPQR STUVWXYZ0123456789 ..)+$*-/, Wt
ABCDEFGHIJKLMNOPQR STUVWXYZ0123456789
ABCL?EFGHIJKLMNOPQR STUVWXYZ0123456789
ABCDEFGHIJKLMNQPQR STUVWXYZ0123456789 .)+$*-/,(='*
ABCDEFGHIJKLMNOPQR STQVWXYZ0123456789 .)+$*-/,(=t$
ABCDEFGHIJKLMNOPQR STUVWXYZ0123456789 .)+$*-/,(_.'$
ABCDEFGHIJKLMNOPQR STUVWXYZ0123456789 .)+$*-/,(='$
ABCDEFGHIJKLMNOPQR STUVWXYZ0123456789 .)+$*-/,(='*
ABCDEFGHIJKLMNOPQR STUVWXYZ0123456789 .)+$*-/,(_'$
ABCDEFGHIJKLMNOPQR STUVWXYZ0123456789 .)+S*-/,(._'t
ABCDEFGHIJKL.MNOPQR ST[OVWXYZ0123456789 .)+$*-/, (=' $
ABCDEFGHIJKLMNOPQR STUVWXYZ0123456789 .)+$*-/,(=~$
ABCDEFGHIJKLMNOPQR STUVWXYZ0123456789?,)+$*-/~(=r*
AI33CDEFGHIJKLMNOPQR STUVWXYZ0123456789-'.)+$*-/,(-'$
AI3CDEFGHIJKLMNOPQR STUVWXYZ0123456789
ABCDEFGHIJKLMNOPQR STUVWXYZ0123456789
ABCDEFGHIJKLMNOPQR STUVWXYZ0123456789,:.,.)+$*-/,(=I$
ABCDEFGHIJKLMNOPQR STUVWXYZ0123456789`'.)+$*-/,(_'$
ABCDEFGHIJKLMNOPQR STUVWXYZ0123456789 .)+$*-/,(_'$
ABCDEFGHIJKLMNOPQR STUVWXYZ0123456789.1 .)+$*-/,(_'$
ABCDEFGHIJKLMNOPQR STUVWXYZQ123456789:','.)+$*-/,(_'$
ABCDEFGHIJKLMNOPQR STUVWXYZ0123456789....)+$*-/,(='$
ABCDEFGHIJKLMNOPQR STUVWXYZ0123456789 )+$*-/,(_'*
ABCDEFGHIJKLMNOPQR STUVWXYZQ123456789 ',_..)+$*-/,(_'$
ABCDEFGHIJKLMNOPQR STUVWXYZ0123456789 :.)+$*-/,(=r*
ABCDEFGHIJKLMNOPQR STVVWXYZ0123456739);',t..)+$*-/,(_'$
A1CDEFGHIJKLMNOPQR STUVWXYZ0123456789 .??..)+$*-/, (=' $
ABCDEFGHIJKLMNOPQR STUVWXYZ0123456789 *
ABCDEFGHIJKLMNOPQR STUVWXYZO123456789;:.,. )+$*-/, $
ABCDEFGHIJKLMNOPQR STUVWXYZ0123456789
ABCDEFGHIJKLMNOPQR' STUVWXYZ0123456789 ;'.)+$*-/,(='$
ABCDEFGHIJKLMNOPQR STUVWXYZ0123456789.i..)+$*-/,(_'$
ABCDEFGHIJKLMNOPQR STUVWXYZ0123456789?:?'.)+$*-/,(='$
ABCDEFGHIJKLMNOPQR STUVWXYZ0123456789
ABCDEFGHIJKLMNOPQR STUVWXYZ0123456789 ?j=J.)+$*-/, (=t$
ABCDEFGHIJKLMNOPQR STUVWXYZ012345.6789-:-;..)+$*-/, (_'$
ABCDEFGHIJKLMNOPQR STUVWXYZ0123456789`;;;.)+$*-/,(_'$
ABCDEFGHIJKLMNOPQR STUVWXYZ0123456789.:'-.)+$*-/,(_'$
ABCDEFGHIJKLMNOPQR STUVWXYZ0123456709 .)+$*-/,(='$
AI3C)DFGHIJKLMNOPQR STUVWXYZ0123456789
A1CDEFGHIJKLMNOPQR STUVWXYZ0123456739...`.)+$*-/,(_'$
ABCDEFGHIJKLMNOPQR STUVWXYZ0123456789i:??)+$*-/,(_'$
ABCDEFGHIJKLMNOPQR STUVWXYZ0123456789 .)+$*-/,(='$
ABCDEFGHIJKLMNOPQR STUVWXYZ0123456739 .)+$*-/,(_'$
ABCDEFGHIJKLMNOPQR STUVWXYZ0123456789. .)+$*-/,(_I$
A1CDEFGHIJKLMNOPQR STUVWXXZ0123456789 .)+S*-/,(_'$
A13CDEFGHIJKLMNQP R U XYZ012 456789 ' .)+$*-/ (_I*
$ @47 7 9 0030024-3
A13CDEFGHIJKLMNQ ve frvjw,j t in I- : C
'V,
ABCDEF
NO I fn
 Approved 11or Releags OVIP v5T i1' `"CI~`-' D 604770A002900030024-3
Agency Visit to
Morning of June 11, 1965
The following detailed agenda is pertinent for this visit.
Introduction
Recap - Ima
ge, P
hoto Digital and now Video
for informa
tion st
orage and retrieval.
Document Systems
The Docume
nt Sys
tem.
and Components
Summary de
script
ion of the modules.
System conf
displays.
igurat
ions - terminals, printers,
Modelling & Simulation
Purpose of
simula
tion, averages, particle flow,
interaction.
Techniques
Simulation
progra
m, measured parameters,
Simulation Analysis and
questions,
Model assu
summa
mption
ry of models simulated.
s, input, delays, functions.
Conclusions
Models #1
-
Assembly
through #4
-
Conclusions
-
Questions
Some Key Inputs for System Definition
- Mechanizing a problem.
? User development of system characteristics.
- List and discuss element a.
Key Inputs'to System Definition
The following are some basic inputs necessary, for system definition.
General - Hours per working day.
System Capacity- Current file size by document size distribution.
Projected yearly growth of file.
- Initial base conversion capacity.
- Input rate per day from zero to let year, 2 N
- Purge rate and criteria of. purge.
- Document size and growth. trends.
Approved. For Release 2005/11/21: CIA-RDP78BO477OA002900030024-3
 Approved For Releas AOWT : A 5PP-A* 770A002900030024-3
Input - Quality identification and measurement - range of sample
documents. Resolution, contrast, background density,
form of input, paper, film, etc. 0
- Indexing status of conversion documents and index plans.
- One or two surface capture per page.
- Surfaces per document - a distribution.
- Range of input quantity per day.
- Tolerable delay between document receipt and capture.
- Preference to batch or interleave.
Output - Form of terminals (Printer/Display)
- Total daily activity per'day for let year (earful - size).
- Total daily activity per day for 2nd year and until mature.
- # terminals in system and distances from central - let year.
- # terminals in system- and distances from central - 2nd year.
- # terminals in system and distances from'central - tjhrough
maturity.
- Request distribution per terminal per day.
- Terminal deviation from normal per day.
- Distribution of output pages (surfaces) per document per
request.
- Request delay distiibution;(s'econds between last output and
next request).
Operational Features
Input request priority levels.
- Output request priority levels.
- Operational hours per day (block sizes).
Approved For Release, 2005/11/21. CIA-RDP78,QQ4770A002900030024-3
 Approved For ReleasCh01T1~I4P~770A002900030024-3
-3-
- Index and addressing support for total system.
- Host CPU and time sharing planning.
- System analysis, usage statistics and report generation.
- Identification of back-up file system.
- Shut down and recovery procedures.
- System maintenance agreements.
- Operating physical environment.
- Operational personnel and training.
Awareness of limiting system parameters.
Approved For Re le se,'20,45111'121': ESA PY8B04770A002900030024-3
 ILLEGIB Approved For Release 2005/11/21: CIA-RDP78BO477OA002900030024-3
Approved For Release 2005/11/21: CIA-RDP78BO477OA002900030024-3