PROJECT ORACLE MANAGEMENT REVIEW MEETING(Sanitized)

• Document Type: CREST
• Collection: General CIA Records
• Document Number (FOIA) /ESDN (CREST): CIA-RDP83T00573R000500080010-1
• Release Decision: RIPPUB
• Original Classification: K
• Document Page Count: 8
• Document Creation Date: December 12, 2016
• Document Release Date: June 3, 2002
• Sequence Number: 10
• Publication Date: November 4, 1975
• Content Type: MFR

lJs- 8c/i
Approved .pr Release 2002/08/15: CIA-RDP83T0Q73R0005,r~ ~'U$0O`="f
4 November 1975
SUBJECT: Project ORACLE Management Review Meeting
The ORACLE Project monthly review was held at the Agency
on 30 October 1975. Present at the meeting were:
The meeting closely followed the agenda (attached) except
four specific problems were discussed in the afternoon. The
summary overview, review of the action items, and the cost
review did not add anything to the information package we
received from the previous week. The hardware and
software reviews are discussed in the following pages.
STATI ITL
ST..TI ITL
ST..TI ITL
Approved For Release 2002/08/15 : CIA-RDP83T00573R000500080010-1
 r+'
Appr'led For Release 2002/08/15 : CIA-RDP83T00573R000500080010-1
Hardware Review
STATINTL I presentation followed very closely the written
report which was submitted the previous week. He noted that
the Channel Simulators which are behind schedule do not STATINTL
TI T~
impact the PSAT. II asked that thel
FA
PSAT which is primarily a hardware test be rescheduled for t t
February 1976, just prior to shipment of the hardware. They
did not give any reasons for this change.
STATINTL reported that continued testing of the hardware
has revealed several deficiencies. He is still generating
engineering changes and has started a second shift operation
to work on the problem of data accuracy.II reported at STATINTL
last month's review that the hardware was placed in a controlled
maintenance environment on 4 August. This seems to be a meaningless
statement; most of the engineering labor is going to tasks
other than maintenance. In any case, it is obvious that at
this time more than a field engineer is required to keep the
hardware operational. It should also be noted that total
hardware labor currently charged to ORACLE is at the same
level as that in August 1974.
Software Review
STATINTL presented the software status review. She
ST.4TII ITL
said has been concentrating on preparing the modules
which make up the system to be used for the Pre-shipment
Acceptance Test (PSAT). The Monthly Technical Progress
Report delivered the week before and the viewgraphs she
presented at the meeting indicated all is well and work
is going pretty much as planned. It was only after questioning
that 0 admitted to being late with the system integration STATINTL
and had incorrectly reported six tasks as complete which
have yet to be finished. It was also noted that the test
jobstream we delivered to them in August has not been run
yet.
STATINTL I has set 3 November 1975 as the day of decision
as to whether or not the IPSAT will be conducted STATINTL
as scheduled during the period 16-26 November. F_ I STATINTL
said there is less than a 50% chance of a go decision. This
contrasts mightily with the written reports we have received
STATINTL
from
Approved For Release 200208/15 : CIA-RDP83T00573R000500080010-1
 Approv,&d For Release 2002/08/15 : CIA-RDP83T00573R000500080010-1
STATINTL I ointed out that it would be absurd to have a
PSAT as scheduled because presently there are restrictions
e-
within the system which will guarantee failure,.c This fstat
adicto
l cont
~
-1
ra
me
byl he would like 0
to run the tests, worry about that time and then
address the contractual problems concerning PSAT.
0 asked us to redefine the tests so II could pass them;
this solution he felt would improve the programmer's morale.
He fears that otherwise the software group will lose its
"momentum". The Agency did not promise to redefine the tests.
After the PSAT discussion,1 reported that PSAT
preparation and training of recently hired programmers has
caused some delays. She also noted that scheduled future
development of the Data Management System is contingent upon
hiring an additional programmer during the first week in
November. Discussion about the Transport Driver Interface
Module which has been scheduled for a rewrite revealed the
work is necessary to permit that software to receive normal
program maintenance. nded her briefing by
describing their newly developed Program Trouble Report system.
, - . - L... T- "Am N anr9 the state of
STATINTL
STATINTL
----
to check the a a and
I PSAT. i immediately called STATI ITL
asked or explanations on all points. He admitted to he
problems but would not discuss reasons or responsibilities.
He said higher level management had not been informed. I
1 at the
b
-
I SULLWdLC ucvc--v +.. --
by Messrs two weeks ago that the project is not
going well. These gentlemen discovered several serious
problems which will be discussed later in this report. TheTTIITL
most astounding was that 0 did not want the government S
d t accuracy of the ORACLE system during the STATINTL
ems
asked him to be prepared to present these pro
Management Review Meeting. By the time of the meetingTTIITL
1 .1- : ~ -1-1 r1 c9, nuious not to test
aspect of PSAT. As to the other problems,0 i
bring them up for discussion; therefore, I asked that theyTTITL
be considered separately later in the da had already SA
been aware of the problems turned up by and STATINTL
. - -- - .c ; r rc -{-hcam TL
L Cd.L1GGU ~++. . -
data accuracy and so dropped their request to delete that
d'd not STATUTE
it is e# ciL Lilcy ILu_ - - --
until the final acceptance period. Based on history I K I I I L
of reticence, I am concerned there could be other serious
problems waiting for us.
Approved For Release 2002/08/15 : CIA-RDP83T00573R000500080010-1
3
 Approved For ReI se 2002/08/15: CIA-RDP83T00573R0O95500080010-1
ST.4TII ITL
ST.4TI ITL
ST.4TI ITL
ST.4TI ITL
ST.4TI ITL
ST.4TI ITL
Major Problems
Expandability of the ORACLE Hardware
informed us that when additional hardware is
added to the system there will be certain restrictions.
They are briefly summarized below:
a) "Perfect switching" is lost between Data Channels (DC's)
and External Data Channel Processors (EDCP's) when the
number of EDCP's is increased.
b) "Perfect switching" is lost between Data Channels (DC's)
and Transport Drivers (TD's) when the number of DC's is
increased and/or the number of TD's is increased.
has defined "perfect switching" as the ability of one
device to be accessed by any other device. Perfect switching
permits straightforward software handlers and guarantees that
any single device failure will not cause the system to be
inoperable but will result only in some degree of degradation.
During the discussion, took the position that STATINTL
they never said perfect switching would be provided with
hardware expansion. At this time software has not been
designed to handle the more severe problems of accessingTTIITL
and recovery from device failures. suggested STATINTL
we document our requirements and wi review them.
Functional Growth of the ORACLE System
revealed that at any given time only one of
the Storage Control Processors (SCP) can access the entire
bank of Transport Drivers. Thus although we have two SCP's,
it becomes impossible to isolate hardware devices such
that production can be serviced with one while local
modifications and major enhancement efforts can be supported
by the second one.
again took the position that they have not committed
t
a design error. As before] I asked that we ocumen
our requirements for review. I _J privately sT..TI ITL
expressed that the reason we purchased two SCP's was out of
generosity.
A side light of this problem is that this situation
inhibits some types of hardware diagnostics. Complete
Approved For Release 2002/08/15 : CIA-Ri)P83T00573R000500080010-1
 Approved For ReI se 2002/08/15 : CIA-RDP83T00573R01 00080010-1
ST.4TII ITL
ST.4TI ITL
ST.4TI ITL
ST.4TI ITL
testing and repair of the Transport Driver Interface can
only be performed by shutting down the production system.
The same holds true for initializing the mass storage tapes.
This operation cannot be performed on the spare SCP. it
must be done using the active SCP which we feel can only be
used to service production.
Software Deficiencies
has designed the ORACLE software such that a
dedicated disk controller is required to permit an SCP
to service the internal system files. This aspect of the
des is contrary to the Design Specification; however,
IIn denies this. They have flatly stated that without a
dedicated controller the system will simply not function.
The result here is that more money and space will be
required for back-up controllers. More importantly, the
design they are proposing introduces a condition where a
single device failure can bring down the entire production
system. If the dedicated controller fails, a spare one must
be switched to the SCP while the failed one is being repaired.
The specification calls for shared controllers such that a
failure will only degrade performance rather than bring the
system to a stop.
Last but not least, 0 stated that the system will
use only one SCP at a time, rather than two. The second SCP
is only for redundancy and will be of use only when the
first one fails. At this time we cannot judge the impact of
this unilateral decision to change the specification. If
one SCP can provide the necessary performance level, no great
harm will be done. Further, the control software will be
greatly simplified. Agency agreement here will provide C
the opportunity to apply manpower to other areas.
My view is that all of the problems described above
result from not adhering to the specifications stated in
the 0 technical proposal and in the 19 March Mass Storage
System Design both of which are part of the contract. All
but the last problem, single SCP control, have adverse
consequences. Most systems suffer from shortcutting but
planned deficiencies for a system that has a strong chance
of being a one-of-a-kind will surely invite disaster. The
software team has already shown it has severe problems with
its performance. I would not like to see that same group try
to mitigate hardware problems with clever software.
Approved For Release 2002/08/15 : CIA-RDP83T00573R000500080010-1
5
T.4TI ITL
 ? Approved For ReIse 2002/08/15 : CIA-RDP83T00573R01500080010-1
Finally, the most serious aspect is that did not
voluntarily inform us of any of the problems. We discovered
them all after the fact.
Manager, Project ORAJ=
ST.4TII ITL
ST.4TI ITL
Approved For Release 2002/08/15 : CIA-RDP83T00573R000500080010-1
 Approved Foor~Release 2002/08/15: CIA-RDP83TO04PR000500080010-1
AGENDA
ST.4TII ITL
TMS-2 MASS STORAGE SYSTEM
OCTOBER PROGRESS'REVIEW
I. LOCATION
C.I.A. Headquarters, McLean, Va.
II. SCHEDULE
Thursday, October 30, 1975, 0900 - 1430 Hour
III. AGENDA ITEMS
0900 - 0915
0915 - 0930
0930 - 1000
1000 - 1015
1015 - 1130
1130 - 1300
1300 - 1315
1315 - 1330
1330 - 1430
Introduction/Summary Overview
Review of Meeting Action Items
Hardware Review/Status
Break
Software Review/Status
Lunch .
Cost/Financial Report Review
Other Open Items/Establish Schedule for
November Review
Summary Review and Action Items List
WMS
10-20-75
Approved For Release 2002108/15 : CIA-RDP83T00573R000500080010-1
 STAT Approved For Release 2002/08/15 : CIA-RDP83T00573R000500080010-1
Next 2 Page(s) In Document Exempt
Approved For Release 2002/08/15 : CIA-RDP83T00573R000500080010-1