REPLACEMENT FOR THE IBM 360/50 TIME SHARING COMPUTER SYSTEM

• Document Type: CREST
• Collection: General CIA Records
• Document Number (FOIA) /ESDN (CREST): CIA-RDP71R00510A000300180009-5
• Release Decision: RIPPUB
• Original Classification: C
• Document Page Count: 5
• Document Creation Date: December 16, 2016
• Document Release Date: December 3, 2004
• Sequence Number: 9
• Publication Date: July 17, 1968
• Content Type: MF

Approved For Rele
rotirlityrN iliAtDP71 R0051 0A0003001,8
AtSC-
MEMORANDUM
THROUGH
SUBIECT
Pu
DD/ S
July 1968
tflrsctar-CornptseUsr
Director for Science and Technology
merit for the IBM 0/50 Time
g Computer System
1. This memorandum contains a request for guidance in para-
3, prior to acquisition of computer equipment.
2. Among the severel computers in OCS is an ISM System 360,
Model SO which was purchased last year with the understanding that it
would be turned over to the Clandestine Services early in calendar
year 1969. It is time now to consider the system to replace that SO in
OM, since IBM must be gives sufficient lead time to ensure delivery
of the replacement system. We have investigated 11 alternatives which
could be considered compatible with our production systems, and have
briefly considered, and rejected, six others which are not.
3. Time-sharing is one of the major capabilities of the current
generation of computers. It allows multiple users at terminalsoshich
can be placed at locations remote from the computer files, to engage
in a ' dialogue with these files -- each user has the impression that he
has the system entirely to himself. His inquiry is answered by a type-
writer printout or a television-terminal.like display almost as fast as
he can enter the question; the response enables him to refine his ques-
tion or seek additional information.
4. There is no doubt in my mind, nor, I hope, in the minds of
others, that CIA ehould have a time sharing capability. The question
s: how much of one, how soon? There are, currently, several good
applications: security name-checking I mathe-
matical computations for OEL and other engineers; on-lino debuffeull at
newly written computer programs; COWS; etc. There are other. softer,
"requirements.' What is certain is that requests will fellow demon-
strated capability. We need to provide for both developmental work and
operational support.
Approved For ReleroKr2L10 LCIA-RDPI1 'RAW a 0018
--MT.77? ?
Excluded from autcr,zitic
ULNIIAL
Doise,T.
5FtJit
25
Approved For Release 2005/02Dan-Alt.P71R00510A000300180009-5
CONFIUL
5. One serious consideration, particularly for COINS, is thecur
ty problem that exists when differed terminals querying &fieriest files
are linked with one computer--or, in a nen-time sharing system, when
more than one program is running an one machine (multitasking). In
either case, it is possible that information from ens filo might be printed
out or displayed, to a customer querying another file. We have pro-
posed a solution which we think will work, but can't claim 100% certainty;
others. including ARPA, are working on the problem. But, at the mom-
ent, the Office of Security feel. constrained to say that when administra-
tively sensitive files or multi-level security files are on a particular
computer, that computer cannot be used by non-Agency people in the
COINS network. nor. In most circumstances, can it be time shared
within the Agency unless the sensitive files are removed.
? 6. A very serious consideration is qualified people -- and enough
of them to ensure that time sharing developments don't occur at the
expense of other operational programs, and modifications thereto. I am
not now coacerned about slots, but the right guy. There just aren't
very many people with the ability to develop time sharing systems -- in
fact, within this Office there is really only one man who understands the
OCR time shaming system in depth, and he is the one who developed it.
Further he has not had time to document what he has built. There are
systems that were developed outside which, thosigh still complex, would
require something less than the unique talent alluded to above -- and
this is a factor to consider even theses most of the non-Agency systems
still have serious drawbacks. Two of direct interest are System Develop
meat Cor oration's ADEPT program, which ORD will be testing. and
Lincoln Lab's Command Program (CP), which looks very good with ths
IBM 360/67 computer.
7. The eleven alternatives considered rang*monthly rental costs
from $37.280 (for an RCA Spectra 70/46) to $66,951 r an IBM 360/67),
with the average around $54,000. The present 360/50 system, as noted,
has been purchased, but a monthly rental figure for it would be *host
$52, 000. No single system, at the moment, would have Security's bless
ing, particularly if used for COINS, unless during prime shift COINS use
all other jobs were removed from the system a costly accommodation
in terms of impact on Computer Center operations.
3. The alternatives with which / am wrestling boil dews in my
mind, to two: the 360/67, or a 360/50 (replacing those items transferred
to DDP plus acquisition of a 360/40 for COWS. The 50/40 complex
2
Approved For Relea.sA. 2005/02/10 : CIA-RDP71R00510A000300180009-5
uuNHDENTIAL
Approved For Release 510A000300180009-5
? 2??Ntn-Ittrrrikto
would rent for $ 8,000 per month. the COWS 40 cos f which would be
about $17,700. My difficulty in choosing between the two stems (rem the
fact that the 360/67 ha. real advantages from a technical standpoint and
dots represent the forward step which we will probably take in about a
year anyway; the more conservative 50/40 alternative is less expensive,
would more closely satisfy Security and is equipment we have used.
9. The 360/67 is working, is available and meets all of our objec-
tives except that of the Office of Security to avoid putting all time-shared
files on one physical piece of equipment. But on this point it has a new
feature which is hard to describe in lay terms. called the virtual machine
concept, in which one machine system can be made to act as if it were
several -virtually separate machine systems, with special hardware
features to protect each such from the others. This capability, which
has not yet been investigated by the Office of Security, has, we feel, a
higher probability of error protection than the electronic commut4cetion
systems now in use with Security approval. The cost of the 67 is within
planned budgetary limits and, though higher, is not disturbing since the
reserve of power could be put to good use in supplementing the normal
job processing (batch processing) equipment already in the Center. Con-
version to the 67 is fairly simple; both experiment/dim and production
can be dont on one machine. But, adequate manpower resources must
be provided for this more complex system and, even now, time sharing
developments *re competing with other teleprocessing work, i.e., we
might have to sacrifice other activities to devote resources to exploit
the 67.
10. The /40 combination I. sal,r and perhaps sails. the man-
power picture, although it doss increase the total number of systems by
one. (It should be noted that if the individual -who developed our 50 sys-
tem should be unavailable for any reason, we would have to backstop
him with others so the net advantage hers is conditioned by our depen-
dant* on one man./ The 50/40 is less expensive -- by about $9000/
month. And, being Use ambitious, more closely jibes with the known
time sharing requirements picture today. (Again noting that OCS objec-
tive. and resources would, in all probability, be the controlling factor
In user requirements definition). On the security side, it provides an
easier solution, but it also, in truth, only puts off the security question--
it does not answer it. Economics would force another look later.
11. U we go with the 50/40 combination, we should review the
button again in the Fall of 1969 or when the fourth 360/65 system (or
Approved For Release 2Ca/NIFIIDEIN-0,071R00510A000300180009-5
Approved For ReleasedffkrtErgr71R00510A000300180009-5
would TO
further dev
ng idered. Tkiould leave some major options
lug the objective of system stability too far. She SO
system for sheid a year, end effort invested to
p services would not be wasted.
12. Among the OCS senior officers tbev is a clear split between
the hardware/software people and the applications people. th? former
(including my Computer Seism* Advisor, the Advanced Project* Staff
Chief and the Director of the Computer Center) favoring the 67. My
deputy, with stated reluctance, suggests the 50/40 to avoid sacrificing
other activities. My intuition says to go 67; my Scotch Presbyterian
heritage says 50/40.
13. Having reduced the considerations to two, I frankly seek
management guidance on Agency time sharing direction, to determine
which of the two systems to procure. Considerable documentation exists
reflecting the investigations conducted and judgements made; these,
obviously, are available should you wish them. 'has a copy 25
of the basic study.,
/s/ Charles A. Briggs
CHARLES A. BRIGGS
Director of Computer Services
Distribution:
Orig & 1 - Addressee
1 - Executive Registry
1 - DDS&T
g DDS&T Registry
2 - OCS (subj and Chrono)
D/OCS:CABriggs:jbs
(17Jul68)
Approved For Release 2e0Nripalik4-1TEPRIIR00510A000300180009-5
IV UM
25
'Approved For Release 2005/02/10 : CIA-RDP71R00510A000300180009-5
Although the above memo is mor* lengthy than / normally
like to forward, I believe it is well that you get this
rather detailed summation, of the problem. I must admit
that this appears to be a real toss-up and I have no
strong feelings as to which is the best course of action.
On balance, I believe that the 360/67 option is the better
choice.
Approved For Release 2005/02/10 : CIA-RDP71R00510A000300180009-5