EPA
Vol. 1
November, 1978
No. 9
FUTURE DATA PROCESSING IN THE EPA
Willis Greenstreet, Director, MIDSD
Before I get into the 1980's ADP Procure-
ment Program, I want to look a moment at
ADP in general today.
Machine speeds are Increasingly deter-
mined by interconnect distances, so that
a four-fold circuit speed improvement
translates, at best, into a two-fold im-
provement in machine speed. To achieve
significant increases in computation
speed and reliability (such as the major
EPA applications require), we must change
the general architecture, and perhaps the
programming and solution structure.
The architecture must be changed to add
more parallelism, but this will mean
higher costs. We must "tune" the machine
structure to the specific EPA problem
solution, and "detune" for general use.
"Tuning" must be interactive with the
problem solvers (the numerical analysts,
physicists, meteorologists, etc.). We
depend upon their special knowledge and
foresight for the solution methodology of
six to ten years hence.
EPA has a great investment in established
methods (file and data formats, compiler
peculiarities, etc.), and is increasingly
precluded from experimenting with a new
system which significantly changes user
methodology. Speed and reliability im-
provements employ more complex assemblies
of hardware and software than those EPA
has currently. The design control meth-
ods must be sophisticated and complete.
This translates into longer development
time, and hence greater costs. The costs
to investigate a new technology may ex-
ceed the total of EPA's past computer
development costs.
To summarize, today the development of
the 1980*3 resources will require a long
time and a close working relationship
between industry and EPA, especially in
the conceptual and developmental stages.
Because of circuit complexities, coordi-
nation problems, and sophisticated design
control methods, this development may
take longer than in the past, and the
cost will be very high. The incentives,
however, have increased for private
enterprise to participate.
The existing U.S. Government procurement
policies and methods for large computers
require competitive bids. The Government
employs a "fly before buy" policy, in
other words, a full-blown demonstration
of the large computer network before they
commit to a purchase.
This requirement seems reasonable, but in
the past the Government approached the
procurement of large computer networks in
the same manner as they did for a mini-
computer. They wanted to buy large com-
puters off-the-shelf. They wanted to
look up the next buy in the GSA schedule.
They wanted to pay a volume production
price for Unit #1. They expected private
enterprise to spend several million dol-
lars or more for a developmental system
just to have the opportunity to qualify
for a bid. Any company that failed to
qualify or qualified and failed to win
was out several million dollars.
1 can assure you that no one in industry
could play this game for very long. For-
(continued on pg. 2, col. 2)
-------
DIRECTOR'S
CORNER
Willis Grccnsircci
A few weeks ago, I addressed the EPA Con-
ference on the 1980's ADP System Acquisi-
tion and Implementation Program (SAIP).
Sam Brown asked me to make some comments
for the kick-off of the conference and a
copy of my introductory remarks is in-
cluded in this EPA Data Ta,lk. The con-
ference introduced the Agency to SAIP.
The concensus was that the emphasis on
reliability and maintainability in the
program was proper. I am confident that
the new government procurement policy for
ADP resources (OMB Circular A-109) will
allow us to obtain highly reliable ADP
resources in the early 1980's.
But what about from now until then?
We have made extensive improvements to
the Univac Data Center which was fraught
with problems in its early years. It is
now operating at a satisfactory or better
level.
The capacity of the IBM Data Center seems
to be adequate since the installation of
the IBM 3032 processor. However, the
usability and user satisfaction of those
resources have yet to reach my expecta-
tions. The Telecommunications Network is
still not fully operational or dependable
even though we're almost two years into
the contract. We still have many dedi-
cated high-speed lines separate from the
Network.
A few months ago, I transferred responsi-
bility for the IBM Data Center/Network to
Sam Brown. He has established a group in
Waterside Mall to manage that contract
and hired Curt Lackey from Region IV as
Project Officer. Since this is currently
our biggest problem area and Sam is per-
sonally spending the majority of his time
on the 1980's program, he has detailed
Don Fulford to Washington to work full-
time on-site at COMNET. Don's main ob-
jective is to perform an overall review
ranging from operations methodology, to
user serviceability, to general manage-
ment . Don will evaluate whether we can
realistically expect any significant im-
provement. If so, he will prepare a plan
to accomplish that and will be given full
authority to execute it. He will also
advise Sam concerning the adequacy
(growth capability and user service-
ability) of these resources to serve our
needs until the 1980's.
SAIP can provide the more reliable re-
sources required in the next decade.
GREENSTREET (cont. from pg. 1)
tunately, our Government has realized
that they will have to support the devel-
opment of large computer networks in the
same way they have supported major
defense industries. For example, our
Government recognizes the prohibitive ex-
pense facing private enterprises in the
development of advanced aircraft. They
further recognize the risk of advancing
the state-of-the-art and the uncertain
success of the final product. Conse-
quently, over a decade ago policies of
"fly before buy" were introduced. This
urge to "kick the tires" before a pro-
duction commitment is desirable and
clearly understood by industry. The
great expense of advanced aircraft
development results in a Government
policy which fully funds multiple sources
during development and operating proto-
type phases. True competition—another
desirable process—occurs in the testing
of the prototypes with the winner re-
ceiving a production contract. He does
so only with moderate risk. But, more
importantly the loser also has faced only
a modest risk. He survives as a business
to compete again and to contribute to the
security of the country through advance-
ments in the state-of-the-art and has a
chance to win the next round.
The Government's new computer procurement
process has borrowed a page from the
books of major weapons system procurement
policy. A phased procurement will be em-
ployed in the development and construe-
-------
tion of future large computer networks.
The procurement will have four phases.
Phase 1f Definition; A number of
parallel system design contracts will be
issued to qualified companies. The goal
will be to clarify and define the re-
quirement through discussions between the
would-be user and the would-be designer.
Phase 2, Design Contracts Option:
Several companies from Phase 1 are funded
with follow-on design contract options.
Designs are specified as completely as
time and resources permit, and the prob-
lems to be solved are similarly defined
more exactly to minimize the risks during
the construction phase. In this phase,
the design must be verified to the great-
est extent possible, using the specified
design criteria. Current simulations
techniques can achieve this.
Pha.se ^, Demonstration: A full de-
monstration of the major prototype for
each final design is performed. Even
though only one design will be selected
by the contracting agency, I suggest that
there will indeed be two winners—the so-
called loser may decide to invest the
necessary funds to complete his new
design for other purposes. In any event,
there certainly will be two complete, new
computer designs.
Phase 4^ jjqpj.ementation: A winning
design is selected and an implementation
contract is awarded. The selected vendor
completes the engineering, builds, and
installs the system.
no great profit in these stages for in-
dustry, but at least the risks are no
longer unlimited, since much of it is
funded by the Government. Thus, high-
technology systems enterprises will
participate.
In summary, times have changed. Major
advances in computer performance and
reliability will no longer come from cir-
cuit improvements alone. As a result, a
large computer network will have to be
"tuned" to the problem in order to
achieve maximum performance and reli-
ability. Past Government procurement
practices were not attractive to the only
resource capable of developing reliable
large computer networks, that is, high-
technology systems enterprises.
The old policies prevented industry from
developing highly reliable technology.
With the new policies, Government and
industry in the U.S. can cooperate and
advance technology for the benefit of
both.
I believe what is needed in EPA's 1980's
program is a dedication to cooperation
between the program manager's office, the
computer industry, and you, the user.
With the expertise we've put in the pro-
gram manager's office and your know-how
in applications, a large computer network
with 100 times the present performance
and reliability is achievable, and the
stage is set for performance even beyond.
This approach differs from the analogy to
the military aircraft procurement in that
"building the hardware system" will in-
clude a great deal of off-the-shelf
equipment. Yet, a configuration for
specific criteria and the building of
special-purpose interfaces requires high
technology systems expertise.
I realize that risks for both the Govern-
ment and the supplier will still exist.
However, by containing the risks, the
main obstacle to attracting the key in-
dustries is removed. There will still be
LIKE TO SEE YOUR NAME IN PRINT?
If you have an article that you think
might be of interest to the Anp com-
munity, or if you have photographs of
interest, please submit then to the
EPA Data Talk office. You will re-
ceive full acknowledgement for any
items used.
-------
1980's SAIP CONFERENCE
Mike Steinacher
The first Agency-wide meeting on the
1980's AOP Systems Acquisition and Imple-
mentation Program (SAIP) was held in
Research Triangle Park, North Carolina,
on October U & 5, 1978. The purpose of
the meeting was to give program manage-
ment and AOP system managers the back-
ground to SAIP, its current status, and
near-term plans. The conference was
heavily attended.
Willis Greenstreet introduced the con-
ference and welcomed Agency personnel.
Sam Brown presented the background and
charter of the SAIP. He then presented
current status and near-term plans. Mike
Steinacher presented an overview of the
acquisition process under the new policy
of OMB Circular A-109. Don Worley pre-
sented some of the detailed considera-
tions in EPA's requirements.
Three projects have been initiated under
the Program. Each of these were pre-
sented by the contractor performing the
project. The Systems Development Corpo-
ration is developing a security/privacy
study and compiling application system
descriptions for a Data Package to be
released to interested contractors. The
Computer Sciences Corporation is per-
forming a documentation assessment on
application systems. An Arthur Young/-
General Electric team is developing
reliability/maintainability factors for a
Life Cycle Cost Methodology which will
heavily emphasize the programmatic impact
of poor data center reliability.
A Management Advisory Committee (MAC)
with general managers at the Division
Director level or above was established
with representatives from across the
Agency. The MAC is composed of:
Sam Brown, Program Manager
Willis Greenstreet, Director, MIDSD
Bill Mathis, Director CMD
John DeFord, Director, OA, RTP
Bill Benoit, Acting DAA, OA
Herb Barrack, Eastern Regional
Representative
Doug Shape, Central Regional Repre-
sentative
Mike Anderson, Western Regional
Representative
Bob Neligan, Management Representa-
tive, Air
Bruno Vasta, Management Representa-
tive, Toxics
Ned Notzon, Management Representa-
tive, Water
Brian Molloy, Management Represen-
tative, Enforcement
Randy Shope, Management Representa-
tive, R&D
Marylou Uhlig, Administrator's Staff
A SAIP Technical Assistance Committee
(TAC) was also established. ADP repre-
sentatives from across the Agency at-
tended and will be the primary group to
assist the Program Management Office in
requirements definition, evaluations,
etc. While continuity of membership is
important, membership in a group this
large will vary throughout the life of
the program. About the only requirement
for membership is full-time permanent em-
ployment at EPA, mainly in ADP. However,
total membership will be controlled by
the Program Management Office so that
activities can be pursued in an orderly
fashion. The TAC is presently composed
of:
Regional ADP Branch Chiefs (Region
VIII absent)
Mike Platt, Personnel System Manager
Sam Conger, Storet System Manager
Jim Haramerle, SAROAD System Manager
Elgin Fry, Pesticides System Manager
Don Thie, Grants System Manager
Tom Martin, Drinking Water System
Manager
Linda Tucker, Mobile Air, Ann Arbor
Jerry Slaymaker, Stationary Air, RTP
Jerry Nehls, R&D, RTP
Bob Browning, R&D, RTP
Mickey Cline, R&D, Athens
Paul Thorpe, R&D, Las Vegas
Bob Andrews, R&D, Duluth
Jon Broadway, Air, Montgomery (ab-
sent)
Jim Chamblee, Water, D.C.
Mitch Cumberworth, Air, Ann Arbor
Rosanne Light, OE, D.C. (absent)
Tony Jover, Toxics, D.C. (absent)
Ed Nime, ADP, Cinn.
Bruce Rothrock, ADP, Cinn. (Absent)
-------
Kent Smith, FMD, D.C.
Edward Milch, Air, D.C.
Additional attendees, other than the SAIP
staff, at the meeting for general assis-
tance purposes were:
Chris Larkin, Budget
Jean Wilkinson, ADP Budget
Morris Yaguda, M1DSD
Ken Byram, M1DSD
Peg Hall, MIDSO
Joe Fallica, Cost Analyst, SAIP
Steve McNamara, Audit, D.C.
Doug Richmond, Contracts, RIP
Jerry Schuller, Contracts, SAIP
Both committees will be formally char-
tered by the Program Management Office as
the acquisition process continues.
The first meeting concerned information/-
status reporting, and that objective
seemed well accomplished.
ADP BARRIERS OVERCOME
Jerry Miller, Region 5
human beings are always about one-
thirtieth of a second behind. It takes
us that long to get sensory data from the
real world to the brain and make informa-
tion of it.
EPA's managers are also always behind.
Events occur with startling rapidity and
the data they generate are often con-
verted to usable information too late for
timely decisions and actions.
We in data processing have spent our pro-
fessional lives attempting, often futile-
ly, to shorten this time lag, to massage
data into information fast enough for it
to do some good.
We are making a little progress. New
programming and systems analysis disci-
plines have helped. 1 think, however,
that new and improved tools are our only
hope.
The PDF-11/70 minicomputer, together with
the INFORM Data Management System, is one
such tool. Excellent tools, as carpen-
ters and mechanics know, apply to limited
situations and problems. A screwdriver
is a poor tool for driving nails. PDP-
11/INFORM is no different. It is effec-
tive only when used for the right appli-
cations, such as small to moderate
tracking systems.
It removes many of the old barriers be-
tween the user and his/her data, and thus
decreases the time lag we are struggling
with.
One barrier which virtually disappears is
computer system unreliability. The hard-
ware and software seldom fail. Another
broken barrier is speed. Most users in-
teract with the system over hard-wired
lines at 9600 baud with excellent re-
sponse time. INFORM then supplies an
immediate window into the data. Yet
another barrier which separates users and
their information is programming. PDP-
11/INFORM does not remove this barrier,
but does significantly lower it. Users
can enter, look at, modify, summarize,
and analyze data with surprising ease.
It's a good tool, as yet underused in our
region. he intend to change that in
coming months.
RELIABILITY •
THE PRIMARY REQUIREMENT
OF 1980's BATA CENTER
Sam Brown, NCG Director
Who would board a large commercial air-
line jet which, despite several hours of
preventive maintenance every week, had an
almost certain probability of a catas-
trophic airborne failure within a few
hours?
Why do we expect so much more reliability
with an airplane than a computer? They
aren't any more expensive! They are not
even kept in a special environment. It's
just as intolerable to allow the collec-
tion of gadgets we call a computer to
-------
possess unreliability as it is to allow
the collection of devices we call an air-
plane to possess unreliability. We ex-
pect perfection in an airplane. Vie allow
a small degree of minor problems such as
poor lighting or air temperature but not
total airplane failure. I expect the
same from the contractors competing to
design and provide our 1980's ADP
resources - the simple perfection of no
system failures.
The most vital requirement in the acqui-
sition will be that the resources con-
tinuously operate even though a component
fails. The priority of this requirement
grew out of the maturing of EPA ADP uses
from heavy batch processing to the in-
creasing interactive operation - which is
becoming wide-spread and will be even
more so in the 1980's with vital programs
dependent upon ADP resources being avail-
able upon demand. ADP system failure in
the 1980's environment will have a devas-
tating impact on EPA program effective-
ness. Because of increasing dependence
of EPA programs on ADP systems, reliabil-
ity is the paramount objective in the
acquisition program.
In EPA's 1980's ADP Acquisition and Im-
plementation Program (SAIP), emphasis
will be centered on three cornerstones in
priority order:
o Maximize Reliability & Maintain-
ability
o Optimize Quality Assurance
o Minimize Total Life Cycle Costs
In support of these areas, long term
action is being taken to modernize and
tailor EPA specifications and acquisition
strategy to ensure reliability. A major
project has been initiated to:
o Make a comprehensive determina-
tion of the reliability and main-
tenance factors which impact the
EPA ADP environment.
o Implement extensive reliability
and maintainability initiatives
to make a huge and noteworthy im-
pact on the ADP acquisition in
order to dramatically improve the
availability and usefulness of
these resources to the EPA
programs.
o Develop a Total Life Cycle Model
with the major emphasis on
reliability.
In government ADP acquisitions conducted
to date, these fundamental steps have
been lacking. Total emphasis has been
placed on computer capacity and lowest
purchase/lease price. A proper acquisi-
tion program is now possible under the
new procurement policies (OMB Circular
A-109) and EPA's acquisition program is
designed to achieve:
o Reliability and Maintainability
o Quality Assurance via Demo (show
me before I buy)
o Minimum Total Life Cycle Cost
(not purchase price)
The three main mechanisms being empha-
sized to accomplish these objectives are:
o Evaluation Criteria driven by
Reliability Factors
o Total Package Testing Before
Award
o Top Management Attention
From the outset of SAIP, with reliability
specifications and contractual require-
ments, it will be necessary for con-
tractors to orient their managerial and
technical organizations to religiously
pursue the reliability objectives. They
will have to enter into a type of
failure-free warranty wherein they will
be financially encouraged to produce the
most reliable system possible and stand
behind it with responsible service. We
will concentrate on reliability at the
start vice trying to develop it after
award. The ability of a corporation to
compete in this new environment will be
measured by the quality of the system
produced and the reliability it demon-
strates in the service environment.
Front-end investment in engineering,
-------
development, demonstration, and logistic
support are a must since we intend to
maximize the gains available in total
life cycle costing of the EPA ADP
acquisition.
Secondly, full prototype testing of the
actual system will be performed before
EPA awards a final contract - no paper
designs will be procured! The testing
will be total, oriented toward EPA's
mission requirements and heavily weighted
on the reliability objectives.
Under this approach, testing will be ex-
amined as a total package with each test
element clearly defined, assign, and in-
tegrated among appropriate activities.
All too often, testing (benchmarking) has
lacked central direction, coordination,
and purpose oriented toward comprehensive
objectives. Integral to the EPA acquisi-
tion process will be a system designed to
ensure that each system discrepancy is
identified, categorized, resolved, and
weighted in selection considerations in
favor of significantly improved
reliability.
Finally, top management attention will be
brought to bear with the Administrator
making decisions at key points through
the acquisition process.
A new day is possible under the new pro-
curement policy of the Office of Federal
Procurement Policy. We're going to
achieve it. We ain't gonna accept medi-
ocrity no more!
USER ORIENTATION SEMINAR
TO BE ENHANCED
John Staley, SDC/ISI User Services
The orientation seminars, begun in July
for new users of the major EPA data cen-
ters, have proved an effective way to
inform the uninitiated about the data
processing services and facilities at
their disposal. However, the presen-
tations have revealed that their infor-
mation needs go beyond the "new" users of
the centers.
Beginning with the seminars at NCC User
Services in RTP on November 11 and at WCC
User Support in Washington's Waterside
Mall on November 21, the orientations
will be presented in a new format. The
morning session will be a broad intro-
ductory overview of both the WCC and the
NCC. The afternoon session will go into
more detail about information needed to
actually run various applications on the
two systems. This format was chosen to
accommodate people with varying degrees
of data processing expertise and
experience. While the morning session
will be useful to anyone who will use the
data centers, those who neither need nor
desire detailed information may choose
not to attend the afternoon session.
MIDSD and the user support organizations
of both WCC and NCC have been pleased to
make the orientations available to the
EPA data processing community. We feel
the planned enhancements will make an ex-
cellent program even better.
For more information about the presenta-
tions and registration, contact Sherry
Hix, NCC User Services, at (919) 541-3618
(FTS 628-3648), or Pam Stephens, WCC User
Support, at (202) 488-5900.
EPA Data Talk la published monthly by
the National Computer Center, Manage-
ment Information and Data Systems
Division, for EPA personnel and con-
tractors interested in general ADP
topics.
Comments and suggestions are soli-
cited and should be addressed to:
William C. Allen
Editor, EPA Data Talk
National Computer Center
Research Triangle Park,
North Carolina 27711
To ensure that our distribution list
is up-to-date, please indicate any
required changes on the mailing label
attached to this issue and mall it to
the above address.
-------
REGIONAL DESK
puters to improve the quality of data in
the national systems.
lack Sweeney
The new fiscal year promises an increas-
ing trend by the Regional Offices toward
the use of minicomputers and computer
center facilities management contracts.
Currently Regions 4, 5, and 10 are suc-
cessfully operating PDP-11/70 minicom-
puters and Regions 8 and 9 have recently
purchased/leased this minicomputer. The
remaining regions are either completing
or have completed feasibility studies
and, if adequate funds are available,
minicomputers will soon be operational in
these regions. The minicomputer is prov-
ing to be a valuable tool in the regional
offices. Its use promises not to under-
mine the large central EPA computers and
national information systems, but rather
to more efficiently use the central com-
The increasing demand for regional com-
puter center services and the inadequate
in-house staffs to meet this demand are
.resulting in facilities management con-
tracts. Currently, Regions 2 and 9 are
employing Computer Sciences Corporation
under the DM&O contract to operate their
computer centers. Regions 1 and 3 will
be implementing similar arrangements this
fiscal year, and several EPA laboratories
have or will implement similar contract
orders. The Region 2 task order includes
seven persons providing keypunching, key-
verifying, computer operations, and pro-
duction control services from 7 a.m. to 7
p.m., plus documentation and basic pro-
gramming tasks. This region has received
quality services under this contract and
recommends this approach to other region-
al offices.
UNITED STATES
ENVIRONMENTAL PROTECTION AGENCY
National Computer Center
Research Triangle Park
North Carolina 27711
OFFICIAL BUSINESS
PENALTY FOR PRIVATE USE S3OO
AN EQUAL OPPORTUNITY EMPLOYE*
POSTAGE AND PEES PAID
U S ENVIRONMENTAL
PROTECTION AGENCY
KPA-339
THIRD CLASS
BLK. RT.
------- |