EXCEPTION STATEMENT ON ADP COMMITTEE REPORT
Document Type:
Collection:
Document Number (FOIA) /ESDN (CREST):
CIA-RDP67B00446R000600050010-0
Release Decision:
RIPPUB
Original Classification:
S
Document Page Count:
20
Document Creation Date:
December 19, 2016
Document Release Date:
August 30, 2005
Sequence Number:
10
Case Number:
Publication Date:
June 11, 1965
Content Type:
STATEMENT
File:
Attachment | Size |
---|---|
![]() | 880.31 KB |
Body:
ATTACHMENT E
Approved For Release 2005/11/21 : CIA-RD 0010-0
11 June 1965
EXCEPTION STATEMENT ON ADP COMMITTEE REPORT
The purpose of this exception statement by the DD/S&T Representative
to the ADP Committee is to sharpen the focus on the nature of the or-
ganizational issue in question here and to provide Agency Management
with some coverage of the economic and the technical considerations
relevant thereto. Also included under "Conclusions" is a course of
action for consideration.
2. The management question posed in Action Memorandum A-444 is:
Given CIA's increasingly heavy investment in
computing activities, how should Agency Manage-
ment organize those resources for the next five
year period to maximize effectiveness at minimum
cost?
Stated in alternative form, the basic question becomes:
Shall CIA continue and extend until complete its
present effort to centralize computing resources
at the CIA level?
Shall CIA dissolve what is now structured at
the CIA level (OCS) and reorganize its computing
resources at the Directorate level?*
The Office of Computer Services (OCS), established in 1963, now
provides computing support to all CIA Directorates. OCS is the
sole computer facility for Headquarters Building support to three
(DD/I, DD/S, and DD/S&T) of the four Agency Directorates. --
DD/P has a separate facility, established prior to the creation of
OCS, which is adjacent to the CIA Computer Center on the Ground
Floor, Headquarters Building. (Outside the Headquarters Building
there is an NPIC computer facility serving NPIC needs
also established prior to the creation of OCS. )
( ~ n r: r I
eIc!cda,' rent 3niasaljC
dc. ~ratiing and
ass!IiCation
Approved For Release 2005/11/21 : CIA-RDP67B00446R000600050010-0
Approved For Release 2005/11/21 : CIA-RDP67B00446R000600050010-0
3. The ADP Committee's majority conclusion is that the CIA-level
centralization thrust, which has characterized CIA organizational
actions in recent years, should not be continued; instead, it should
be checked... and even reversed. The Committee's report is in-
tended to lay the groundwork for a) assuring continued separateness
of the Agency's present computer centers, and b) creating a manage-
ment climate within which additional centers might reasonably be
expected to emerge.
4. There are three logical areas of consideration regarding the or-
ganization of CIA's computing resources: (a) technical, (b) economic,
and (c) political. The Committee has seriously concerned itself in
both its discussions and its report with but one of these areas: the
political.
5. If it is reasonable to assume that Management, before making this
study assignment to the Committee, was already familiar with the
major political considerations involved here,then what Management
most needs is coverage of the other aspects of the question; i. e. ,
the economic and technical considerations. The Committee has not
provided coverage of same to Management in its report because it
was the majority opinion of the Committee that sufficiently reliable
technical and economic projections were not now possible as
ingredients to Management's decision on the organization of CIA
EDP** resources. The DD/S&T Representative, however, feels
economic and technical projections not only can be made but, indeed,
must be made if Management is to be equipped to consider this or-
ganizational question. These are the sine qua non ingredients of
such decision.
The term "political" is not used to deride or belittle this area of
consideration. Political considerations (such as present organi-
zational lines, functions, habits, preferences, and plans) are
important and must be taken into account by Management in any
organizational decision. Directorate security concepts, as a
specific, are part of these "political" considerations. (See
Para. 8 below)
The acronym "EDP" (Electronic Data Processing) is used in this
report to mean "computing" or "relating to computers.
01
S
Approved For Release 2005/11/21 CL4RDP67B00446R000600050010-0
Approved For Release 2005/11/21 : CIA-RDP67B00446R000600050010-0
6. TECHNICAL CONSIDERATIONS:
a. State of the EDP Science:
The computing science is moving into a dramatic third
generation. New generation capabilities in both hardware
and software constitute the largest step forward in both
productivity and reduced costs in the history of the com-
puting era. As a consequence, the new generation capa-
bilities will have a major impact on all aspects of computing,
including the organization of computing resources.
Some new generation characteristics:
- New generation gear is faster by several
orders of magnitude.
- Its high speed memories have much larger
capacities.
Immediate "on-line" access to very large
information files is now feasible (several
billion characters of on-line storage).
A vast array of remote terminals for input,
query, and display can now be handled... a
requirement in upcoming computing support
to all CIA Directorates.
New generation systems will operate under
sophisticated Executive Control Programs
which enable the computer to pace and control
its own work... to the near exclusion . of the
Computer Operator in the traditional sense.
Manual control of the computer from job to
job (serial processing) will be grossly
inefficient.
The capacity for work is increased manyfold.
Thruput for the IBM System 360 Mod 67, for
example, is equivalent to eight 7090's or
thirty 1410's or thirty-five RCA 501's. . . even
when operated serially.
-3-
Approved For Release 2005/11/21 : CIA-RDP67B00446R000600050010-0
Approved For Release 2005/11/21 : CIA-RDP67B00446R000600050010-0
The larger models of the new generation computers
provide major bonuses over the smaller models. in work capacity, in sophistication of techniques,
and in significantly lower costs per unit of data
processed.
b. Impact on Organization:
The impact of the above processing capacities of the new
generation hardware and software will be felt in virtually all
aspects of computing--management, user, and EDP areas alike.
The direction of impact on organizational structure is
patently clear. The tremendous work capacity, the far-reaching
systems design implications, and the pronounced cost advantages
of the larger models relative to the smaller models will force
greater centralization throughout the computing field.
c. The new generation OCS Computer Center:
During the past twelve months, OCS has developed its plan
to shift to new generation equipment as rapidly as possible. All
computers in the present CIA Computer Center will be replaced
by but two IBM 360/Mod 67 computers. The first one will be
installed in nine months; the second twelve months thereafter.
CIA will be one of the first major centers to shift to new genera-
tion capabilities. This is much more than a status symbol; the
production and economy gains are truly remarkable.
Some specifics on the OCS third generation center:
- The OCS configuration will be a twin-Mod 67
system to which the full range of peripheral
devices can be attached as required by Agency
needs.
The OCS system is a complete "fail-soft" system.
That is, all components are duplicated in the system
so that failure of any component does not stop the
functioning of the system... it only reduces thruput.
The on-line systems now in development in support
of all Agency Directorates will require such fail-
soft support. (This fail-soft structuring of the
machine configuration will tend to characterize
future computing centers of whatever size. )
-4-
Approved For Release 2005/11/21 : CIA-RDP67B00446R000600050010-0
Approved For Release 2005/11/21 : CIA-RDP67B00446R000600050010-0
The specific configuration of hardware selected
by OCS is based on the OCS workload as OCS is
now charged within CIA (all Headquarters com-
puting except RID. )
Projected scientific computing needs of themselves
would require a system of similar scale to that
planned for the CIA Computer Center because of the
stringent speed, memory, and backup requirements
which characterize such applications. The production
capability of the system, however, would be only very
partially tapped if limited to scientific applications
alone.
The Mod 67, selected for the CIA Computer Center,
falls in the "upper middle" size range. There are
two larger models and three smaller models announced
to date. (The Mod 67 is the smallest model of the new
IBM 360 series which possesses the I/O controller and
sub-selector multiplexor capabilities. These capa-
bilities, in themselves, represent a major third
generation stride.) The next smaller model (Mod 50)
does not have the capacity to support the Agency's
more demanding scientific tasks... and is considerably
more costly per unit of information processed.
No change in the main frame configuration of the OCS
Center would be required to do all Headquarters Building
computing as anticipated for the next 3 to 5 years.
If and when additional main frame processing power
were needed at some future date, a third CPU could
be very simply added to the planned configuration at
a cost of about $17, 000/month. This additional CPU,
at $17, 000/month, would provide additional processing
power to the system equivalent to five Mod 501s. each
one of which would cost much more.
Approved For Release 2005/11/21 : CIA-RDP67B00446R000600050010-0
"'~'
Approved For Release 2005/11/ w n DP67B00446R000600050010-0
OCS estimates that all Headquarters Building
computing (OCS and RID) can be done on the
Mod 67 system. within a single shift. This is
intended as a measure of work volume only.
Any center, of course, which serves remote
terminals in user areas must remain open
during periods of authorized usage.
d. Other pertinent specifics:
New generation gear will process most efficiently
with a job mix of maximum diversification. If re-
stricted, organizationally or otherwise, to a single
application, much of its thruput power is wasted.
The concept of re-structuring Agency EDP resources
along Directorate lines suggests the following penalties:
1) The Mod 67, while doing DD/S&T processing,
could concurrently process all other Agency
computing with less than 15% delay (under worst
peak conditions) to DD/S&T processing.
2) All non-DD/S&T internal computation tasks would
be serviced by the Mod 67 in less than half the time
required on Directorate centers equipped with their
own Mod 50's (the next size down from the Mod 67)
and all purely input/output flows (no computation)
would run on the Mod 67 system in slightly better
time.
Software extensions and improvements over time will
continue to increase the thruput of the gear under dis -
cussion so that appreciable increases in workload can be
absorbed without acquisition of additional main frame
equipment. OCS has projected a utilization of 130 hours
per month initially on the Mod 67. Software improvements
should permit-the handling of a four-fold increase in work
in something less than double the machine time.
Approved For Release 2005/11P67B00446R000600050010-0
Approved For Release 2005/11/21 : CIA-RDP67B00446R000600050010-0
SECRET
Can the IBM 360 software, which is yet to be proven in
operations, be counted on to meet specifications? - OCS
feels the risk here is a safe one. IBM has traditionally
equipped its users with the strongest software packages
in the computer industry. This massive corporation has
staked its entire future on the success of the 360 system--
including software. IBM is less likely to fail in its com-
mitments than its users are to fail to fully exploit the
capabilities being offered.
The full exploitation of new generation capabilities will
require on the part of user organizations advanced tech-
nical capabilities in the systems programming area.
Although OCS has made a strong start over the past two
years at assembling a pool of such skills, system pro-
gramming skills will be in short supply in CIA and
throughout the computing industry with the advent of
third generation processing. Decentralization of com-
puting to the Directorates would serve to make the
shortage of such hard-to-find and slow-to-grow skills
all the more critical.
The Committee's report contains the following statement:
if... 'on-line' applications involving complex interplay
between manual and computer based reference facilities...
are most effectively managed under complete Directorate
control. "
The above statement is in no way concurred in
by the DD/S&T representative. There are no
technical arguments to support it,. and it is
directly contrary to good economics.
This section presents some measure of the price tag attached
decentralizing Agency EDP resources to the Directorate level.
hree types of resources have been considered: a) machine rental
' Lars; b) people; and c) Headquarters floor space. The figures
give are minimum figures in each instance. Although this section
on economics is brief, its relevance and import to the organizational
issue is second to none.
Maior economic overtones are apparent in the preceding section on 'Dkporaimd RdnRe1dase3L2096b11/21.:tiCIA-RDP67B00 treREV
050010-0
-7
Approved For Release 2005/11/2 c4 67B00446R000600050010-0
In measuring the hardware rental costs for additional
centers, the following guidelines were followed:
(1) Costs of "remote on-line" devices and extra
capacity peripheral equipments (such as mule:le
data cells) were excluded. (Basically, they are
constant costs whether hooked to Directorate
Centers or to the CIA Center and are thus irre-
levant to Management as a decision factor).
(L) Only those peripheral devices necessary to the
oper._on of a stand-alone center have been in-
cluded.
(3) A double -thruput configuration has been assumed
for all centers because:
EDP systems in development to support
all Directorates (CHIVE, MIS, scientific
systems, WALNUT, etc.) call for remote
terminals in user areas. The machine
language data store for such systems is
retrievable _ay with great difficulty in
the event of equipment failure. Thus, in
practice, it would be highly improbable
i::at any center, centralized or decentralized,
would not find valid justification for double
-chruput support within the first two years of
the five year period under consideration.
The minimum additional annual rental for equipment for any
Direccora--tto- eve1 center is presented in Table 1 on the next
page.
Approved For Release 2005/11/IAP67B00446R000600050010-0
25X1 Approved For Release 2005/11/21 : CIA-RDP67B00446R000600050010-0
Approved For Release 2005/11/21 : CIA-RDP67B00446R000600050010-0
Approved For Release 2005/11/2k -, 4 67B00446R000600050010-0
b. Additional People:
The personnel figures given in Table 2 are, in each case,
minimums extra personnel requirements to be expected with the
endorsement of Directorate -level centers instead of CIA-level
centralization.
Computer Cperators - Multiple computing centers, of
course, mean more computers (2 per enter) and more
computers mean more Computer Operators. Two-shift
operations are assun.ed in Table 2.
Systems Programmers - As pointed out in Paragraph 6 ,
above, third generation gear and remote terminal systems
will force the development and use of Executive Programs
for automatic control of operations. Systems Programming
will become a must in staffing any center. Our figure of
four Systems Programme- s per local co ,?_ er is an initial
level only; expansion to more than twice r_