Feasibility Study of an EPA
Data Communications Network
Phase II Report
Information & Communication Applications Inc.
-------
C-74-672
Feasibility Study of an EPA
Data Communications Network
Phase II Report
Submitted to:
Theodore R. Harris, Computer Specialist
Technical Project Officer
Management Information and Data Systems
Division (PM-218)
Environmental Protection Agency
Washington, D .C . 20460
RFP No. WA-74-E127
January 27, 1975
INFORMATION & COMMUNICATION APPLICATIONS,INC.
Suite 710 - 6110 Executive Boulevard, Rockville, Maryland 20852
(301)770-4100
-------
TABLE OF CONTENTS
Page
1. INTRODUCTION AND EXECUTIVE SUMMARY 1-1
1.1 Background 1-1
1.2 Phase II 1-2
1.3 Phase II Conclusions and Recommendations 1-4
1.4 Planning Recommendations 1-10
1.4.1 Utilization Statistics 1-13
1.4.2 New Common Carrier Offerings 1-14
1.5 Summary 1-16
2. DATA COMMUNICATIONS REQUIREMENTS 2-1
2.1 Current Requirements and Facilities 2-1
2.1.1 Major Resource Providers 2-2
2.1.2 Resource Users 2-2
2.2 EDP Planning 2-4
2.2.1 Short-Term Planning 2-4
2.2.2 Long-Term Planning 2-6
2.2.2.1 User Structure 2-6
2.2.2.2 Computer Facilities 2-8
2.2.2.3 FY 75 and FY 77 Communications Requirements 2-8
3. DESIGN ANALYSIS 3_!
3.1 Design Constraints 3-1
3.2 Assumptions 3-2
3.2.1 Uniformity of Terminal Speed 3-2
3.2.2 Division of Workload Between Washington and RTP 3-3
3.2.3 Importance of Front-End Resource Unavailability 3-4
-------
TABLE OF CONTENTS
Page
3.3 Input Data 3-5
3.3.1 Load Location Data 3-5
3.3.2 Cost Parameters 3-8
3.3.3 Usage Distribution 3-17
3.4 Software Functions 3-18
3.4.1 Utilization Reporting 3-18
3.4.2 Network Integrity and Security 3-20
3.4.3 Error Reporting for Networks and Terminals 3-20
3.4.4 Information Routing 3-21
3.4.5 Speed and Transmission Code Recognition 3-21
3.5 Commercial Carrier Evaluation 3-23
3.5.1 DATRAN 3-23
3.5.2 MCI 3-25
3.5.3 Tymshare and General Electric 3-26
3.6 Government Carrier Evaluation 3-26
3.6.1 FTS 3-26
3.6.2 ARPA 3-28
3.6.3 INFONET 3-32
3.6.4 ADMTS 3-33
3.7 Selected Design Techniques 3-33
3.7.1 WATS 3-33
3.7.2 Multiplexors, Concentrators, and Front-Ends 3-35
3.7.2.1 Multiplexors 3-35
3.7.2.2 Concentrators 3-37
3.7.2.3 Front-End Controllers 3-41
-------
TABLE OF CONTENTS
Page
3.7.3 Leased Lines 3-43
3.7.4 Foreign Exchange (FX) Service 3-45
3.8 The EPA PLANET System 3-46
3.9 Design Alternatives I, II and III 3-50
3.10 Alternative Analysis 3-51
4. RECOMMENDED NETWORK - ALTERNATIVE #4 4_j
-------
LIST OF TABLES
Table 1-1.
Table 1-2.
Table 1-3.
Table 2-1.
Table 2-2.
Table 2-3.
Table 3-1 .
Table 3-2.
Table 3-3.
Table 3-4.
Table 3-5.
Table 4-1.
Table 4-2.
Table 4-3.
Table 4-4.
Table 4-5.
Table 4-6.
Table 4-7.
Table 4-8.
Table 4-9.
Table 4-10.
Table 4-11.
Page
Network Cost Justification 1-8
Monthly Comparison of Network Costs 1-9
Monthly Comparison of Network Communication
Costs 1-9
EPA Data Summary 2-9
PLANET Input Data 2-13
Assumed EPA Low Speed and Medium Speed
Terminal Usage Distribution 2-26
Front-End Loading and Data Transfer Estimates 3-53
WATS Only/WATS-MUX Relative Cost Comparison 3-55
EPA Front-End Estimated Port Hardware Costs 3-56
Incremented Hardware Costs 3-58
Alternative Cost Summation 3-59
Communications Network: FY75/FY77 Cost
Summary 4-2
Communications Network: FY75/FY77, NCC
Loading and Data Transfer Estimates 4-3
Local Telephone Service for EPA Computer Site
at Washington, D.C. and RTP 4-4
FY75 Composite Listing Communications Services
for EPA Network - WATS 4-5
FY77 Composite Listing Communications Services
for EPA Network - WATS 4-6
WATS Summary for FY75 4-7
WATS Cost Summary for FY77 4-8
FY75 Composite Listing Communications Services
for EPA Network - TDMS 4-9
FY77 Composite Listing Communications Services
for EPA Network - TDMS 4-10
EPA Cities with Multiplexor Port Expansion
Capacity 4-11
Network Communications Controller Port
Assignments . 4-12
-------
LIST OF FIGURES
Page
1-1 EPA Network WATS Telephone Service 1-5
1-2 FY 75 EPA Network Multiplexor Placement 1-6
1-3 FY 77 EPA Network Multiplexor Placement 1-7
1-4 FY 75 Network Configuration 1-11
1-5 FY 77 Network Configuration 1-12
3-1 WATS Network Hardware 3-9
3-2 Multiplexed Network Hardware and Associated
Cost Parameters 3-11
3-3 Estimated Low Speed Multiplexor Costs 3-13
3-4 Estimated Low Speed Multiplexor Costs
(High Density Cities Requiring Additional Capacity) 3-15
3-5 Estimated Medium Speed Multiplexor Costs 3-16
3-6 Example of Packet Switching Network 3-30
3-7 Typical TDM Equipment Configuration 3-38
3-8 Concentrator Network Equipment Configuration 3-40
3-9 Front-End Communications Controller Equipment
Configuration 3-44
3-10 Interaction of PLANET Program Components and Data Base 3-48
3-11 EPA Alternative Network Configurations 3-52
-------
1. INTRODUCTION AND EXECUTIVE SUMMARY
-------
1. INTRODUCTION AND EXECUTIVE SUMMARY
The electronic data processing requirements of the Environmental Protection
Agency (EPA) have been growing extremely rapidly. EPA's current data pro-
cessing and communications environment has resulted from the necessity to meet
those requirements. To improve this situation and also provide for more orderly
future growth, the Management Information and Data Systems Division (MIDSD)
began a number of feasibility and planning studies to aid in the definition of a
data processing and communications capability that would provide maximum bene-
fit to EPA users at a reasonable cost.
1.1 BACKGROUND
One study, "Feasibility of Establishing an EPA Data Communications Network",
was conducted by Information and Communication Applications, Inc. (ICA) to
determine the feasibility of an EPA network, and provide EPA with alternative
system designs. ICA analyzed three (3) different communications system struc-
tures, determined that a consolidated network was cost effective and thus
feasible, and developed the least expensive design for the consolidated network.
During this study, many methods of communication system design were analyzed,
including ARPA, INFONET, TYMESHARE, Concentrators, Multiplexors, etc.. It
was determined that a combination of multiplexors with dedicated point-to-point
lines and WATS telephone service would best fulfill the EPA requirements.
The recommended network structure (referred to as Alternative #1) was a single,
nationwide network centered in Washington, D.C., because computer analysis
of EPA's projected communications load showed that it would be the most cost
effective to service this nationwide load by a single consolidated communications
network. Network communications controllers whose functions include gathering
network utilization data and providing code and transmission speed conversion
1-1
-------
capabilities were located in Washington, D .C . In order to provide complete
monitoring and control of the entire EPA network, local data from the Research
Triangle Park (RTP) users was routed to Washington before being input to the RTP
computer systems.
The ICA Phase I final report, "Feasibility of Establishing an EPA Data Communica-
tions Network, " September 18, 1974, contains supportive data from which the
recommended network was derived. It has been revised as described below in
Phase II and is therefore encompassed in this final report, "Feasibility Study of
an EPA Data Communications Network".
1.2 PHASE II
ICA's Phase I final report contained a number of recommendations, the most perti-
nent being:
More extensive utilization data needed to be
gathered, especially in regards to RTP.
Further analysis should be conducted specifically
examining the possibility of using a programmable
communications controller (concentrator) in RTP to
handle local traffic and transfer data to and from the
Washington, D.C. facility.
Upon receipt of this final report, MIDSD personnel pursued these two recommenda-
tions and concluded that:
A Network Communications Controller (NCC) be placed
in RTP in addition to the NCC(s) in Washington, D .C .
to avoid the following problems:
1-2
-------
1. Less reliable communications when local RTF
users are routed through the Network Communica-
tions Controller (NCC) in Washington and then
back to RTF.
2 . Since the location of the Washington Computer
Center is not fixed, the service to RTF could be
severely impacted when the WCC is relocated.
This relocation may occur several times in the
life of the network.
The utilization data for RTF be re-examined as:
1. Local RJE utilization was based on old IBM 360/50
statistics and is estimated to be 50 percent below
the current RJE utilization experienced on the
Univac 1110.
2 . The RTF low speed local utilization projections for
both FY75 and FY77 are estimated to be approximately
50 percent low, based on current/RTP experience.
EPA then suggested an alternative approach to the recommended solution (referred
to as Alternative #4) described as follows:
1. Locate Network Communications Controllers (NCC) at
both Washington and RTF Resource Locations - not
necessarily with equal capability.
2. All users except RTF local users will access the
Washington NCC and specific requests for RTF
resources will be "relayed" to the RTF NCC; other-
wise requestor will access Washington resource.
1-3
-------
3. RTF local users with specific requests for the
Washington resource will be "relayed" to the
Washington NCC and then to the Washington resource;
otherwise requestor will access the RTF resource.
Additionally, based on the RTP-Univac capabilities, it was recommended that the
number of local ports at RTF be increased substantially.
Based on EPA's conclusions, ICA has reanalyzed its Phase I results and developed
a network configuration for Alternative #4. This work is identified as Phase II.
The Phase II network configuration is documented in this report.
1.3. PHASE II CONCLUSIONS AND RECOMMENDATIONS
Alternative #4 provides all remote users access to both Washington, D.C», and
RTF, N.C., EDP resources by a combination of WATS telephone service (See
Figure 1-1) and time division multiplexors connected to Washington, D .C . by
dedicated point-to-point lines (See Figures 1-2 and 1-3). EPA users in Wash-
ington, D .C. and RTF metropolitan areas are provided access to both the Wash-
ington and RTF facilities by local telephone service. The recommended network
is based on a 99 percent EDP resource availability, thus ensuring that the EPA
users will not receive more than one busy signal in every 100 calls. Forty per-
cent of the data received at the Washington central site is transferred to RTF by
high speed data channels.
In performing the Phase II Analysis, ICA determined the overall network costs for
Alternative #4. The cost effectiveness and feasibility of the recommended
network was conclusively determined by a comparison of costs for the FY75 data
load. This comparison shows that the recommended network will save about
$35,424 per month. (See Table 1-1)
1-4
-------
I
Ui
Borion
hester
Narragansett
Band 1
Band 2
Band 3
Band 4
Band 5
Figure 1-1. EPA Network WATS Telephone Service
-------
bps
'4800 bps
7200 bps
9600 bps
i9600 bps
Computer-to-
Computer Link
Figure 1-2, FY 75 EPA Network Multiplexor Placement
-------
X
.Narracanset
2400 bps
4800 bps
7200 bps
9600 bps
9600 bps
Computer-To-Computer Link
Figure 1-3. FY 77 EPA Network Multiplexor Placement
-------
Table 1-1. Network Cost Justification
Existing Network
Recommended Network
(Alternative #4)
Savings Per Month
Projected Monthly Costs'
$149,672
114,248
$35,424
Table 1-2 shows the cost comparison of the recommended network for FY75 and
FY77.
Thus, the feasibility and cost effectiveness of a consolidated EPA communications
network remain unchanged through FY77. In addition, because all data passes
through either the Washington NCC or the RTP NCC, the network control, moni-
toring, and utilization statistics gathering objectives are fulfilled by Alternative
#4*.
To implement the FY75 EPA communications network requires a single network
communications controller (NCC) in Washington, D.C., and a single NCC in RTP.
The NCC in Washington, D .C. should be augmented by an additional NCC in the
FY.77 network, with the channels split as evenly as possible. The FY75 network
will have 228 low speed communications channels (front-end processor ports) and
58 medium speed channels in Washington, B.C., and 60 low speed channels and
These cost projections are based on the FY75 data load and do not include the
cost of network communications controllers (NCC). The NCC costs ($12 ,000/mo.)
are not included because the cost of the line controllers are not included in the
existing network costs. Note that the NCC costs will be higher than the corres-
ponding costs of the existing line controllers ($10,200/mo.) but the NCC
functional capabilities are also higher.
* In Alternative #1, local RTP data could not be directly input to the RTP computer
Computer Systems without seriously violating these objectives. This forced the
Alternative #1 costs to be greater than the Alternative #4 costs.
1-8
-------
Table 1-2. Monthly Comparison of Network Costs
FY75
FY77
Existing Network
$149,672
Unknown
Alternative #4
$114,248
$145,694
Monthly Savings
$35,424
Unknown
Table 1-3. Monthly Comparison of Network Communication Controllers
FY75
FY77
Existing Network
Line Controllers
$10,200
Unknown
Alternative #4
$12,000
(1-Washington ,
1-RTP)
$16,000
(2-Washington ,
1-RTP)
Monthly
Additional Costs
$1,800
Unknown
1-9
-------
11 medium speed channels in RTF. The FY77 network load requires 318 low
speed and 79 medium speed channels in Washington, B.C., and 100 low speed
and 18 medium speed channels in RTF. The cost of the two NCC's for the FY75
network is approximately $12,000 per month. The three NCC's for the FY77
network are estimated to cost $16,000 per month. These costs may vary depending
on the final configuration and vendor selected. They also exclude the costs of
any peripherals needed to augment the NCC's. The cost of the existing line con-
trollers at both computer centers is $10,200 per month. Table 1-3 shows the
monthly cost comparison of the line controllers and the FY75 and FY77 NCC
configuration.
The network configuration for FY75 is shown by Figure 1-4 and for FY77 by Figure
1-5 .. The projected monthly cost for the FY77 EPA network exclusive of Network
Communications Controller costs is $145,694 per month.
1.4 PLANNING RECOMMENDATIONS
Comprehensive utilization statistics are required to determine precise EPA commu-
nications requirements and assure EPA of the most cost-beneficial network that
can effectively handle the changing EPA user requirements through FY77 and beyond,
To ensure early and orderly acquisition of such a network, EPA must immediately
begin concentrated research and analysis of the following areas:
a. Utilization Statistics
b. New Common Carrier Offerings
This research and analysis will further augment the planning recommendations
presented by this study and will provide EPA with the detailed technical knowledge
necessary to implement the most effective and cost-beneficial communications
network. .
1-10
-------
RTP_
Medium
Speed
(11
Channels
Low
Speed
(60
:hannels
u/
CITIES SERVICE!
Remote EPA Multiplexed Users
Seattle (2)
Chicago (2)
Atlanta (1)
Corvallls, Ore
Las Vegas (1)
Athens, Ga. (1)
Now York City (1)
Boston (1)
Dallas (1)
Philadelphia (1)
(0
ru:s SEKVICI
CITH:S s
\
KVICED
Sc-iiUlc (1)
Denver (2)
Chicago (1)
San Francisco (1)
Portl.ind, Ore. (1)
Corvallls, Ore. (1)
Las Vegas (1)
Athens, Ga. (1)
Hrossc lie, Mich. (1)
New York City (1)
Boston (1)
Namagnnsett, R.I. (1)
Rochester, N.Y. (1)
Kosovllle, Minn. (1)
Dall.is (1)
Madison. Wls. (1)
Jackson, Miss. (1)
Kansas City, Mo. (1)
RTF
NCC
(Front End)
9GOO Baud
Jata Transfer
Channels (9)
\VCC
Network
Communl cations
Controller
(Front-End)
Medium
Speed
(21
Channels
Low
Speed
(20
Channels
REMOTE
U
S
WATS
Telephone Service
Medium
Speed
(15
Channels
Low
Speed
(66
Channels
RTP
CPU
WCC
CPU
Local
Telephone Service
Figure 1-4. FY 75 Netorork Configuration
-------
ro
P.TP
S
Medium
Speed
(IS
Channels
low
Speed
(100
Channels
Remote EPA Multlplcxd Users
cirics|ii:Rv;ccp crqi:s SI:RVIC/L>__
CITIKS SI:KV;CI
CITJKS
grnvicr
Seattle (-1)
Denver (1)
Chic.iqo (4)
Atlanta (2)
Corvallls, Oro. (4)
Cincinnati (1)
Athens, Ga. (1)
C.rossc He. Mich. (2)
New York City (1)
Boston (1)
Dallas (2)
Philadelphia (1)
9600 Bjud
Channels (21)
Denver (1)
Chicago (2)
Athens, Ga. (1)
ilio.'ioe He, Mich.
Philadelphia (1)
(1)
7200 Baud
Channels (2)
San Francisco (1)
Portland, Ore. (1)
Corvallls, Ore. (1)
Las Vegas (1)
Ulchfinlfi, Minn. (1)
New York City (1)
Boston (1)
Narnagansett, R.I. (1)
Rochester, N.Y. (1)
Columbus, Ohio (1)
Rosevllle, Minn. (1)
Madison, Wls. (1)
Gulf Breeze, Fla. (1)
Jackson, Miss. (1)
Kansas City, Mo. (1)
4800 Baud
Channels (6)
2400 Baud
Channels (IS)
Netwoi I;
Communications
Controller
(Front -I'.nd)
RTP
NCC
9600 Baud Data Transfer
Channels (13)
\
NctworK
Communications
Controller
(Front-End)
;hannels
Low
Speed
(24
Channels
V/ATS
Telephone Service
Low
Speed
(89
Channels)
Local
Telephone Service
CPU
RTF
CPU
Figure 1-5. FY 77 Network Configuration
-------
1.4.1 Utilization Statistics
Communications utilization statistics are essential in planning and implementing
well-organized and efficient communications networks. Modifications to the
structure of the network to increase its efficiency are determined from them.
Additionally, utilization statistics aid in future network planning by identifying
network growth. Utilization monitoring can be done on individual users, major
locations, and the network as a whole. EPA must study two major areas when
investigating the overall area of utilization statistics:
Methods and procedures for gathering and reporting
utilization data
Continual and regular evaluation of the utilization data
gathered to identify the effects of the changes in utili-
zation on the network design.
The methods and procedures for gathering and reporting network utilization data
are the critical factors in ensuring that accurate data is gathered and that cost-
effective networks are designed to meet the utilization requirements. Continual
and regular evaluation of utilization data allows identification of specific changes
in network structure which when implemented provide a constant and defined grade
of service at the lowest cost.
Gathering and monitoring the utilization statistics for the proposed network must
be planned carefully. Procedures should be established to define precisely the
data to be gathered, and how and when it should be processed. Accurate utiliza-
tion data can only be gathered through such pre-established and standardized
procedures. Since it is very desirable to have the EPA communications front-end
provide pertinent communications utilization data, steps should be taken
immediately to define the utilization data gathering and reporting functions, and
1-13
-------
include them in the RFP for acquisition of the network. Precise functional defi-
nition of utilization data gathering and reporting requirements requires detailed
knowledge of communications networks and systems and of the methods and
modeling tools that use the data to perform network planning analysis.
Utilization of EDP resources should be evaluated regularly. This regular evalua-
tion will ensure that as the network utilization increases (or decreases), the
network design may be augmented by additional equipment and channels that
ensure that EPA users are provided with a constant grade of service (99 percent
system availability).
1.4.2 New Common Carrier Offerings
Several new common carrier offerings appear to provide cost effective alternatives
for the future EPA network. However, because the facilities supporting the new
common carrier offerings are not completely operational, and because the network
recommended by this study must be capable of immediate implementation, only
some of the new common carrier offerings have been recommended for possible use
in the FY75 and FY77 EPA network. All new common carrier offerings that advance
current data communications technology should be continually evaluated for future
EPA network availability. The most promising new offerings are:
a. Specialized Common Carriers (MCI and DATRAN)
b. Digital Data Service (DDS) being developed at AT&T
c. Value Added Networks (Packet Communications, Inc.
and Telenet)
The opportunity exists to obtain private lines from both of the operating specialized
common carriers, MCI and DATRAN. Although neither can serve all of the cities
1-14
-------
requiring multiplexors, both carriers can provide channels to many of the cities
EPA must serve. Because their private line offerings are operational at Wash-
ington, D.C., are cost-competitive, and have attractive error rate guarantees,
they warrant immediate consideration for use in the EPA network.
AT&T recently filed a new Tariff 267 for Digital Data Service (DOS) that provides
for private line services at transmission rates meeting EPA requirements, and at
very competitive costs. DOS is designed specifically for binary data transfer.
The schedule for DDA availability calls for 96 cities by the end of 1976. All EPA
multiplexor locations will be served under this tariff, by that time, except the
following:
Corvallis, Ore. Narragansett, R.I.
Athens, Ga. . Roseville, Minn.
Gulf Breeze, Fla. Jackson, Miss.
Richfield, Minn.
DBS, in some form, is not precluded for all of these locations, and further inves-
tigation should be pursued by EPA. Thus, DBS warrants continued evaluation to
assess its future applicability to the EPA network.
Packet switching, a relatively new concept in data communications, is offered by
the Value Added Carriers. It appears to be an ideal method for handling many
widely dispersed network users. Some commercial packet switching common
carriers have filed tariff applications, but all of them have yet to begin operation.
When these commercial value added networks are operational, packet switching
could be a cost-effective method of structuring the EPA communications network.
However, two areas warrant detailed evaluation before packet switching can be
recommended for the EPA network:
1-15
-------
Response time performance of packet switching networks
servicing low and medium speed terminals
Network costs for servicing widely dispersed low and
medium speed terminals
Some data that can be used to evaluate value added network costs is available
now, but accurate evaluation of response times is not possible because the data
needed cannot be obtained. Because the value added networks appear to offer a
unique and probably cost-competitive service, with very attractive error rate
guarantees, they should be immediately evaluated for potential cost effectiveness
and continually evaluated for their ability to provide responsive network operation.
Therefore, ICA recommends that EPA continually monitor and evaluate the applica-
bility of new common carrier offerings to its network operation.
1.5 SUMMARY
Alternative #4 will, in the long run, provide EPA with the best overall communica-
tion network, but again, in EPA's best interests, ICA recommends in this report
that functional interface and communications requirements be developed for both the
NCC's.
Section 4 of this document describes this revised communications network (Alter-
native #4) that should be procured as part of EPA's data processing services. The
description will include FY75, FY77 cost summary, Network Communications Con-
troller loading, data transfer analysis, and a detailed configuration description
for the EPA network.
Sections 2 and 3 of this report are excerpted from the Phase I report to provide a
continuity for the logic presented within this executive summary. Section 4, as
stated above, has been revised to reflect Alternative #4.
1-16
-------
2. DATA COMMUNICATIONS REQUIREMENTS
-------
2 . DATA COMMUNICATIONS REQUIREMENTS
This section describes EPA current and planned data communications
requirements.
2.1 CURRENT REQUIREMENTS AND FACILITIES
EPA is responsible for all of the major federal programs dealing with air and
water pollution, pesticide regulation, environmental radiation, and solid
waste disposal. Thus/the information processing requirements are many
and diverse. EDP activities must support scientific and administrative
functions in addition to building, maintaining, and accessing large data
bases. Meeting these requirements is made even more difficult because
users are widely dispersed geographically. Further, state and local
governments must be supplied with limited EDP support. As a result, EPA
has had to provide its users with nationwide computer and communication
facilities that can support both conversational and remote batch access
modes.
At its formation, EPA inherited a goodly number of on-going programs and
responsibilities. At the same time, the EDP resources it inherited were
.far too small to handle its EDP requirements. Further, Agency headquarters
in Washington, D. C., the focal point for many major programs, was with-
out major EDP resources of any kind. Therefore, as a matter of expediency,
EPA entered into a number of interagency agreements and contracts with
private industry for EDP support. The present EDP environment results
from these conditions.
2-1
-------
2.1.1 Major Resource Providers
EPA now uses the services of several outside government agency and
private industry EDP resources, plus its own facility in Research Triangle
Park (RTP), North Carolina. The two major EDP outside suppliers are
National Institutes of Health (NIH) and Optimum Systems Incorporated (OSI).
Both are in Washington, D. C. OSI is by far used the most. This facility,
which handles the vast majority of workload directed to Washington, B.C.,
consists of an IBM 370/155 and 370/158 with 192 ports for both low
(75-1800 baud) and medium (2000-9600 baud) speed utilization. The major
systems software used by EPA are HASP, WYLBUR, and TSO.
The major IBM 360/370 systems software used by EPA at the NIH facility
are WYLBUR, CPS, HASP, and TSO. In addition, EPA also uses the NIH PDP-10
computer to some extent on scientific applications.
The RTF facility consists of a UNIVAC 1110 with a total of 64 ports for both
low and high speed utilization. Since this facility has been in operation
only since the fall of 1973, utilization and user demands on it have yet to
be determined.
2.1.2 Resource Users
The present EDP structure consists of three major segments, the Washington,
D. C. area, the RTP area, and the remainder of the nation. Washington, D. C.
supplies most of the EDP services. It suppots the headquarters administrative
and financial requirements, and handles most of the EDP requirements of the
Water programs within the Office of Water and Hazardous Material (OWHM).
RTP provides most of the EDP services for the Air programs within the Office of
Air and Waste Materials (OAWP). In addition, both Washington, D. C. and
2-2
-------
RTF must supply EDP services to the ten EPA regions, NERC's and various
laboratories scattered around the nation.
At the present time, EPA users outside the Washington, D. C. area, but
serviced from there, are supported with a communications network that OSI
has planned specifically for them at a cost of approximately $80,000 per month.
This communications network consists of leased lines with multiplexors,
dedicated lines, and WATS lines. The multiplexors are in the following cities.
Athens, Ga. (SERL) Kansas City, Mo. (Region VII)
Boston, Mass. (Region I) Portland, Ore. (NERC)
Palo Alto, Calif. (Region IX) Philadelphia, Pa. (Region III)
Chicago, 111. (Region V) Dallas, Tex. (Region VI)
Atlanta, Ga. (Region IV) Denver, Colo. (Region VIII)
Seattle, Wash. (Region X) Cincinnati, Ohio (NERC)
This communications network supports the various cities with a heavy concen-
tration of EPA users. EPA users not supported by this network use the
Washington, D. C. resources via the Federal Telephone Service (FTS) or the
switched network.
The RTF facility only provides minimal direct network support for EPA users
outside its area. Consequently EPA users outside the RTF area who desire to
use the RTF resources must do so via FTS, WATS lines, or the switched net-
work. Due to the lack of information and the heavy use of FTS, it is not
possible to give a reasonable cost estimate for RTF's present communications
cost.
2-3
-------
Since the number of EPA users requiring EDP support has been growing ex-
tremely rapidly, EPA is taking a number of steps intended to provide a high
grade of service at a reasonable cost.
2.2 EDP PLANNING
EPA is presently planning both short-term and long-term solutions to the
problem of supplying convenient and economical EDP resources.
2.2.1 Short-Term Planning
OSI is planning to improve its existing communications network, and RTF is
planning to establish a communications network to support its remote users.
OSI plans to introduce an independently programmable COMTEN communica-
tions front-end to its two IBM 370 systems. In addition, OSI plans to upgrade
its multiplexors to recognize 300-baud terminals and below automatically.
This will allow OSI to take advantage of the software speed recognition in-
herent in the COMTEN and provide better service for EPA's many different
terminal types. Finally OSI plans to increase the transmission capacity of
their multiplexors and leased lines to 9600 baud. The cost to EPA for these
improvements is yet to be determined.
RTP's immediate plans call for the installation of four multiplexor lines that
will provide service to Washington, D. C., Philadelphia, Pa., New York, N.Y.,
and Boston, Mass. The equipment to be installed, and the RTF communica-
tions support that will result in these four EPA regions are shown below.
i
A. WATS Lines
Four groups of two lines AREA 6 INWARD:
2-4
-------
(1) 1 full period and 1 measured on a rotary for the
IBM 2741 type terminals (15 cps)
(2) 1 full period and 1 measured on a rotary for the TTY
type terminals (30 cps).
(3) 1 full period and 1 measured on a rotary for the
UNIVAC 1004 type terminals (2000 baud).
(4) 1 full period and 1 measured on a rotary for the
IBM 2780 type terminals (2000 baud).
B. Leased lines with multiplexors
Four leased multiplexed lines as follows:
(1) Research Triangle Park to Washington, D. C. (4800 baud)
(2) Research Triangle Park to Philadelphia, Pa. (4800 baud)
(3) Research Triangle Park to New York, N.Y. (4800 baud)
(4) Research Triangle Park to Boston, Mass. (4800 baud)
This configuration supports one RJE (2000 baud), one TTY terminal (30 cps),
and one IBM 2741 type (15 cps) in each city and allows for future expansion.
RTP will provide similar communications support to the remaining six EPA
regions in the near future. The estimated cost to EPA for the complete net-
work is approximately $33,000 per month.
The OSI and RTP networks will provide major EPA users with convenient and
reliable access to both the Washington and the RTP resources. However, it
is not clear that these two networks will solve all EPA communications prob-
lems, as the EDP requirements of EPA's users may quickly become larger than
2-5 ' .
-------
the two networks can handle. In addition, having two parallel networks may
not be the most cost effective method in which to service these EDP require-
ments. Therefore EPA is implementing the two networks, and concurrently is
planning for its long-range EDP requirements, of which this study is a part.
2.2.2 Long-Term Planning
EPA is analyzing its computer requirements, terminal requirements (both
conversational and batch), and communications requirements. This study
is primarily concerned with the EPA communications requirements, but it
will be affected by the results of other EPA studies and decisions affecting
the EDP environment. Certain EDP requirements, however, can be readily
seen and must be taken into account in long-term communications planning.
2.2.2.1 User Structure
EDP users can be divided into four categories or levels.
Level I - These are in the Washington, D. C. area and
use their EDP resources extensively.
Level II - These users are in the Research Triangle Park, N.C.
(RTF) area. They also use their EDP resources
heavily.
Level III - These remote users use both the Washington, D. C.
resource and the RTF resource, but their usage is
somewhat less than Level I and Level II.
Level III users are in the following cities:
Boston, Massachusetts (Region 1)
New York, New York (Region 2)
Philadelphia, Pennsylvania (Region 3)
2-6
-------
Atlanta, Georgia (Region 4)
Chicago, Illinois (Region 5)
Dallas, Texas (Region 6)
Kansas City, Missouri (Region 7)
Denver, Colorado (Region 8)
San Francisco, California (Region 9)
Seattle, Washington (Region 10)
Montgomery, Alabama
Athens, Georgia
Annapolis, Maryland
Ann Arbor, Michigan
Grosse He, Michigan
St. Louis, Missouri
Edison, New Jersey
Las Vegas, Nevada
Cincinnati, Ohio
Ada, Oklahoma
Corvallis, Oregon
Narragansett, Rhode Island
Level IV - These users, which are not part of EPA, consist
of state and local environmental boards, and
use both EDP resources, but very infrequently.
These users are widely dispersed throughout
the United States.
This user structure dictates a communications network that can provide ex-
tremely high grade service for Level I and Level II users, high grade service
for the Level III users, and the ability to service Level IV users as the need
arises. Further, the communications network must be able to service both
2-7
-------
conversational (low speed) and remote batch (medium speed) terminals within
all four user levels .
2.2.2.2 Computer Facilities
EPA users are concentrated around Washington, B.C. and RTF since extensive
EDP resources are provided for these two areas. Further, since EPA presently
has computer facilities in both Washington, D. C. and RTF, and plans to
continue both facilities for at least the next five years, the communications
network must be able to support both areas. If more than one EDP vendor is
supporting EPA in the Washington, B.C. area when the communications net-
work is being implemented, only the selected prime vendor should be tied into
the network. It would most likely be cost-effective to have any EPA user re-
quiring service of other resource providers to do so directly, as these require-
ments should be rare.
2.2.2.3 FY 75 and FY 77 Communications Requirements
The FY 75 and FY 77 communications requirements are based on the most cur-
rent data available. The raw data used in this study has been extracted, aug-
mented, and condensed into three tables for use as input to the PLANET modeling
program.
Table 2-1 is a summation of the actual and projected utilization data gathered
from EPA. It should be noted that since the UNIVAC 1110 has been installed
at RTF for a relatively short period, the utilization data for RTF is based on
the EPA IBM 360/50 statistics.
Table 2-2 is the data used for input to the PLANET program, which is the vehi-
cle used by ICA for the modeling of the EPA network alternatives .
2-8
-------
Table 2-1. EPA Data Summation
Location
LEVEL I
Washington. D.C.
LEVEL II
Research Triangle Park, N.C.
LEVEL HI
Boston. Mass. (Region 1)
New York. New York (Region 2)
Philadelphia, Pa. (Region 3)
Atlanta, Georgia (Region 4)
Chicago, Illinois (Region 5)
Dallas, Texas (Region 6)
Kansas City, Mo. (Region 7)
Denver, Colo. (Region 8)
San Francisco, Calif. (Region 9)
Seattle, Wash. (Region 10)
Montgomery, Alabama
Athens, Georgia
Annapolis , Maryland
Ann Arbor, Michigan
Grosse lie, Michigan
St. Louis, Missouri
Edison, New Jersey
Las Vegas, Nevada
Cincinnati, Ohio
Ada, Oklahoma
Corvallls, Oregon
Narragansett, Rhode Island
LEVEL IV
Little Rock, Arkansas
Phoenix, Arizona
Los Angeles, California
Table 2
Equipment
Inven orv
M
6
4
1
1
1
2
2
2 -
1
1
1
2
-
5
1
-
2
-
-
-
6
2
1
1
-
-
-
L
286
60
8
7
14
23
13
5
8
11
S
10
-
9
1
-
1
-
1
' -
29
2
7
-
-
-
-
Table 7
NIH/OSI
Utilization
M
980.58
82.33
167.53
150.77
143.28
135.05
245.23
233.27
102.20
101.03
34.78
327.60
-
207.67
79.23
' -
70.07
-
-
45.02
89.20
.45
58.80
26.60
. -
-
-
L
6641.13
244.28
i
64.18
163.87
354.34
597.63
241.07
35.76
166.76
174.67
200.37
248.97
39.18
124.05
5.63
2.90
112.88
. -
.45
109.21
359.95
7.28
45.71
41.25
11.70
41.83
3.55
RTP1
Average
Utilization
M
21.29
414.72
.50
2.45
3.37
7.16
47.43
9.12
7.68
2.31
42.12
14.16
-
-
-
-
-
-
-
-
-
-
-
-
-'
-
L
.98
1079.18
12.50
10.07
12.09
10.94
5.83
3.20
2.92
4.90
7.15
9.02
.-
-
-
-
-
-
-
-
'-
-
-
-
-
_
Table 3
Workload Committee Projections
FY74 FY77
M
330.3
545.0
160.0
118.0
88.0
272.0
380.0
247.0
107.0
119.8
-
402.0
-
27.0
-
. -
110.0
-
-
140.0
136.0
5.0
225.0
42.0
-
-
~
L
2763.0
855.0
59.5
102.0
602.0
717.0
599.0
106.0
-
438.6
-
452.0
-
300.0
-
-
165. 0
-
-
57.0
987.0
20.0
225.0
102.5
-
-
"
M
734.7
690.0
160.0
209.0
140.0
500.0
981.0
277.0
53.0
195.5
-
770.0
-
45.0
-'
-
510.0
-
-
133.0
180.0
21.0
230.0
60.0
-
-
"
L
5433.2
1772.0
101.6
199.0
889.0
1178.0
1381.0
243.0
-
563.9
-
943.0
-
743.0
-
-
820.0
-
-
179.0
1162.0
41.0
235.0
136.0
-
-
"
Projected Load
(Hours Connect Time/Month)
FY75 FY77
M
1001.87
545.00
168.03
153.22
146.65
272.00
380.00
247.00
109.88
119.80
76.90
402.00
-
207.67
79.23
-
110.00
-
-
140.00
136.00
5.00
225.00
42.00
-
-
"
L
6641.13
1323.46
76.68
173.94
602.00
717.00
599.00
106.00
169.68
438. CO
207.52
452.00
39.18
300.00
5.63
2.90
165.00
5.00
.45
109.21
987.00
20.00
225.00
102.50
11.70
41.83
3.55
M L
1406.27
690.00
168.03
244.22
198.65
500.00
981.00
277.00
55.88
195.50
96.13
770.00
-
225.67
99.04
-
510.00
2.00
-
133.00
180.00
21.00
230.00
60.00
-
-
9311.33
2240. 4',
118.78
270.94
889.00
1178.00
1381.00
243.00
212.10
563.90
259.40
943.00
48.98
743.00
7.04
3.63
820.00
15.00
.56
231.21
1162.00
41.00
235.00
136.00
14.52
52.28
4.43
-------
Table 2-1. (Cont.)
to
I
Location
Sacramento, California
Santa Ana, California
Yreka, California
Hartford, Connecticut
Wethersfleld, Connecticut
Dover, Delaware
Gulfbreeze, Florida
Tallahassee, Florida
Boise, Idaho
Harrlsburg, Illinois
Springfield, Illinois
Evans villa, Indiana
Indianapolis, Indiana
Des Molnes, Iowa
Topeka, Kansas
Frankfort, Kentucky
Louisville, Kentucky
Baton Rouge, Louisiana
New Orleans, Louisiana
Hubbardston, Massachusetts
Ironwood, Michigan
Lansing, Michigan
Duluth, Minnesota
Ely, Minnesota
Richfield, Minnesota
Roseville, Minnesota
Jefferson City, Missouri
Jackson, Mississippi
Lincoln, Nebraska
Bismarck, North Dakota
Trenton, New Jersey
Santa Fe, New Mexico
Table 2
Equipment
Inventory
M
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
' -
-
-
-
'-
-
-
-
-
-
-
-
-
-
~
L
- -
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
- .
**
Table 7
NIH/OSI
Utilization
M
3.25
6.73
-
.37
78.95
-
-
96.52
-
-
-
-
-
-
-
2. 95
1.83
-
-
-
- '
-
-
-
-
-
21.20
-
-
-
L
30.40
-
.68
-
-
'4.88
29.35
40.18
14.10
1.21
8.28
37.60
9.38
.18
34.85
6.21
.68
14.43
4.80
.01
1.50
. 35.56
.05
2.91
52.63
275.58
7.31
59.91
.36
10.50
1.48
7.18
RTF*
Average
Utilization
M
-
-
'-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
"*
L
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
"*
Table 3
Workload Committee Projections
FY74 FY77
M
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
"*
L
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
*"
M
-
-
-
-
-
-
-
-
-
-
-
.-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
**
L
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
- -
-
-
-
-
-
~
Projected Load
(Hours Connect Tine/Month)
FY75 FY77
M
3.25
6.73
-
.37
78.95
-
-
96.52
-
-
-
-
-
-
-
-
2.95
. 1.83
-
- -
-
-
-
-
-
-
-
21.20
-
-
-
~
L
30.40
-
.68
-
-
4.88
29.35
40.18
14.10
1.21
8.28
37.60
9.38
.18
34.85
6.21
.68
14.43
4.80
.01
1.50
35.56
.05
' 2.91
52.63
275.58
7.31
59.91
.36
10.50
1.48
7.18
M
4. 00
8.41
-
.46
98.68
-
-
120.65
-
-
-
-
-
-
-
-
3.68
2.28
-
-
-
-
-
-
-
-
-
26.50
-
-
-
""
L
38.00
-
.85
-
-
6.10
84.35
50.22
17.62
1.51
10.35
47.00
11.72
.22
43.56
7.75
.85
18.03
6.00
.01
1.87
44.45
.06
3.63
65.78
344.47
9.13
74. 8S
.-!5
13.12
1.85
8.97
-------
Table 2-1. (Cont.)
Location
Taos, New Mexico
Albany, New York
Buffalo. New York
Oswego, New York
Rochester, New York
Scarsdale , New York
Cleveland, Ohio
Columbus, Ohio
Oklahoma City, Oklahoma
Portland, Oregon
Chadds Ford, Pennsylvania
Harrisburg, Pennsylvania
Lewlstown, Pennsylvania
Meadvllle, Pennsylvania
Pittsburgh, Pennsylvania
Reading, Pennsylvania
Columbia, South Carolina
Pierre, South Dakota
Chattanooga, Tennessee
Nashville, Tennessee
Austin, Texas
Baileys Cross Roads, Virginia
Elklns, Virginia
Huntlngton, Virginia
Richmond, Virginia
Montpeller, Vermont
Olympla, Washington
Tacoma, Washington
Madison, Wisconsin
Milwaukee, Wisconsin
Charleston, West Virginia
Wheeling, West Virginia
Table 2
Equipment
Inventory
M
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
*"
L
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
2
Table 7
NIH/OSI
Utilization
M
-
1.30
-
-
-
-
-
6.90
.45
-
-
-
-
-
-
-
75.22
-
-
-
-
-"
-
S.10
-
-
4.77
.38
6.58
1.08
-
~
L
20.53
.11
6.01
.53
70.81
8.00
35.35
58.58
40.13
191.35
20.85
36.08
2.30
9.45
6.35
25.01
18.00
8.38
1.46
15.08
1.83
2.38
.68
6.48
14.45
12.33
4.25
-
111.27
10.33
19.92
2.43
RTP1
Average
Utilization
M
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
~
L
' -
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
~
Table 3
Workload Committee Projections
FY74 FY77
M
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
"*
L
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
*
M
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
.-
-
-
-
-
-
-
-
-
-
-
-
~
L
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
~
Projected Load
(Hours Connect Time/Month)
FY7S FY77
M
-
1.30
-
-
-
-
-
6.90
.45
-
-
-
-
-
-
-
75.22
-
-
-
-
-
-
5.10
-
-
4.77
.38
6.58
1.08
-
~
L
20.53
.11
6.01
.53
70.81
8.00
35.35
58.58
40.13
191.35
20.85
36.08
2.30
9.45
6.35
25.01
18.00
8.38
1.46
15.08
1.83
. 2.38
.68
6.48
14.45
12.33
4.25
-
111.27
10.33
19.92
2.43
M
-
1.62
-
-
-
-
-
8.62
.56
-
-
-
-
-
-
-
94.02
-
-
-
-
-
-
6.37
-
-
5.96
.47
8.22
1.35
-
L
25.66
.13
7.51
.65
88.51
10.00
44.18
73.22
50.16
239.18
26.06
45.10
2.87
11.81
7.93
31.26
22.50
10.47
1.82
18.85
2.28
2.97
.85
, 8.10
18.06
15.41
5.31
-
139.08
12.91
24.90
3.03
CO
I
-------
NOTES TO TABLE 2-1
The figures in this column are based on the RTF Operations and
Systems Reports for November, 73 through June, 74 and on the
raw low speed utilization data for RTF contained in Communication
Network Design by William J. Douglas.
2
This data is for the Washington, D. C. metropolitan area and
includes usage from Bethesda, Rockville, Silver Spring, Alexandria,
Annandale, Arlington, and McLean.
3
This data represents the usage for the Research Triangle Park area
and includes network utilization from Raleigh, Durham, and Chapel Hill,
2-12
-------
Table 2-2. Planet Input Data
LOCATION
LEVEL I
Washington, B.C.
LEVEL II
Research Triangle Park,
North Carolina
LEVEL III .
Boston, Massachusetts - Region 1
-
New York, New York - Region 2
Philadelphia , Pennsylvania -
Region 3
AREA
CODE
202
919
617
212
215
TIME
ZONE
E=l C=2
M=3 P=4
1
1
1
1 ,
1
COORDINATES
TARIFF #264
V
5622
6331
4422
4997
5251
H
1583
1499
1249
1406
1458
NUMBER
OF
TERMINALS
286
6
323
12
60
4
116
10
8
1
32
3
7
1
29
3
14
1
40
4
LOAD/WEEK
(CHARACTERS)
166028250
500935000
232783250
703135000
33086500
272500000
56011500
345000000 '
1917000
84015000
2969500
84015000
. 4348500
76610000
6773500.
122110000
15050000
73325000
22225000
99325000
RATE
(CPS)
30
600
30
600
30
600
30
600
30
600
30
600
30
600
30
600
30
600
30
600
YEAR
1=1975
2=1977
1
1
2
2
1
1
2
2
1
1
2
2
1
1
2
2
1
1
2
2
-------
Table 2-2. (Cont.)
LOCATION
LEVEL III (Cont.)
Atlanta , Georgia - Region 4
Chicago, Illinois - Region 5
Dallas, Texas - Region 6
Kansas City, Missouri - Region 7
- . . .
Denver, Colorado - Region 8
San Francisco, California -
' Region 9
-
AREA
CODE
404
312
214
816
303
415
TIME
ZONE
E=l C=2
M=3 P=4
1
2
2
2
3
4
COORDINATES
TARIFF #264
V
7260
5986
8436
7027
7501
8492
H
2083
3426 '
/
4034
4203
5899
8719
NUMBER
OF
TERMINALS
23
2
36
3
13
2
33
4
5
2
33
4
8
1
30
3
11
1
32
4
.
.5
1
LOAD/WEEK
(CHARACTERS)
17925000
136000000
29450000
250000000
14975000
190000000
34525000
490500000
2650000
123500000
6075000
138500000
4242000
54940000
5302500
27940000
10965000
59900000
14097500
97750000
5188000
38450000
RATE
(CPS)
30
600
30
600
30
600
30
600
30
600
' 30
600
30
600
30
600
30
600
30'
600
30
600
YEAR
1=1975
2=1977
1
1
2
2
1
1
2
2
1
1
2
2
1
1
2
2
1
1
2
2
1
1
-------
Table 2-2 . (Cont.)
LOCATION
LEVEL III (Cont.)
San Francisco, Calif. - Reg. 9
Seattle, Washington - Region 10
Montgomery, Alabama
Athens, Georgia
Annapolis, Maryland
.
Ann Arbor, Michigan
-Grosse Isle, Michigan
-. .
AREA
CODE
206
205
404
301
313
313
TIME
ZONE
E=l C=2
M=3 P=4
4
2
1
1
1
1
COORDINATES
TARIFF #264
V
6336
7692
7130
5555
5602
5536
H
8896
2247
1948
1519
2918
2828
NUMBER
OF
TERMINALS
30
4
10
2
29
4
1
9
' 2
9
5
22
4
1
1
5
2
1
26
2
1
2
7 .
3
LOAD/WEEK
(CHARACTERS)
6485000
48065000
11300000
201000000
23575000
471500000
979500
1224500
2500000
7500000
103835000
18575000
112835000
140750
39615000
176000
49520000
72500
90750
5000000
4125000
55000000
20500000
255000000
RATE
(CPS)
30
600
30
600
30
600
30
30
600
30
600
30
600
30
600
30
600
30
30
600
30
600
30
600
YEAR
1=1975
2=1977
2
2
1
1
2
2
1
2
2
1
1
2
2
1
1
2
2
1
2
2
1
1
2
:2
-------
Table 2-2. (Cont.)
LOCATION
LEVEL III (Cont.)
St. Louis, Missouri
Edison, New Jersey
Las Vegas, Nevada
'
<,
Cincinnati, Ohio
Ada, Oklahoma
Corvallis, Oregon
AREA
CODE
314
201
702
513
405
503
TIM£
ZONE
E=l C=2
M=3 P=4
2
1
4
1
2
4
COORDINATES
TARIFF #2 64
V
6807
5069
8665
6263
8029
7016
H
3482
1429
7411
2679
4176
8991
NUMBER
OF
TERMINALS
1
5
1
1
4
1
1
1
26
4
29
6
42
7
2
2
7
3
7
1
30
2
LOAD/WEEK
(CHARACTERS)
125000
375000
1000000
11250
14000
2500000
2730250
70000000
5780250
66500000
24675000
68000000
29050000
90000000
500000
2500000
1025000
10500000
5625000
112500000
5875000
115000000
RATE
(CPS)
30
30
600
30
30
600
30
600
30
600
30
600
30
600
30
600
30
600
30
600
30
600
YEAR
1=1975
2=1977
1
2
2
1
2
2
1
1
2
2
1
1
2
2
1
1
2
2
1
1
2
2
-------
Table 2-2. (Cont.)
LOCATION
LEVEL III (Cont.)
Narragansett, Rhode Island
LEVEL IV
Little Rock, Arkansas
Phoenix, Arizona
Los Angeles , California
Sacramento, California
.
Santa Ana , California
Yreka, California
Hartford, Connecticut
i
AREA
CODE
401
501
602
213
707
714
916
203
'J.'1M£
ZONE
E=l C=2
M=3 P=4
1
2
3
4
4
4
4
1
COORDINATES
TARIFF #264
V
4623
7721
9135
9213
8304
9267
7631
4687
H
1176
3451
/
6748
7878
8580
7798
8841
1373
NUMBER
OF
TERMINALS
1
1
4
2
1
1
2
2
1
1
3
1
3
1
1
1
1
1
1
1
LOAD/WEEK
(CHARACTERS)
2562500
21000000
3400000
30000000
292500
365500
1045750
1307000
: 88750
110750
760000
1625000
950000
2030000
3365000
4205000
17000
21250
185000
230000
RATE
(CPS)
30
600
30
600
30
30
30
30
30
30
30
600
30
600
600
600
30
30
600
600
YEAR
1=1975
2=1977
1
1
2
2
1
2
1
2
1
2
1
1
2
2
1
2
1
2
1
2
-------
Table 2-2. (Cont.)
LOCATION
LEVEL IV (Cont.)
Wethersfield , Connecticut
Dover, Delaware
Gulf Breeze, Florida
Tallahassee, Florida
Boise, Idaho
Harrisburg, Illinois
Springfield, Illinois
Evansville, Indiana
Indianapolis, Indiana
AREA
CODE
203
302
904
904
208
618
217
812
317
TIME
ZONE
E=l C=2
M=3 P=4
1
1
1
1
3
2
2
2
1
COORDINATES
TARIFF #264
V
4687
5429
8153
7877
7096
6851
6539
6729
6272
H
1373
1408
2183
1716
7869
3142
3513
3019
2992
NUMBER
OF
TERMINALS
1
1
1
1
3
3
1
1
1
1
1
1
1
1
1
1
2
2
1
1
LOAD/WEEK
(CHARACTERS)
39475000
49340000
122000
152500
733750
2108750
1004500
48260000 '
1255500
60325000
352500
440500
30250
37750
207000
258750
940000
1175000
234500
293000
RATE
(CPS)
600
600
30
30
30
30
30
600
30
600
30
30
30
30
30
30
30
30
30
30
YEAR
1=1975
2=1977
1
2
1
2
1
2
1
1
2
2
1
2
1
2
I
2
1
2
1
2
-------
Table 2-2. (Cont.)
LOCATION
LEVEL IV (Cont.)
Des Moines, Iowa
Topeka, Kansas
Frankfort, Kentucky
Louisville, Kentucky
Baton Rouge, Louisiana
,
New Orleans, Louisiana
Hubbardston, Massachusetts
Ironwood, Michigan
i '
AREA
CODE
515
913
502
502
318
504
617
906
TIME
ZONE
E=l C=2
M=3 P=4
2
2
1
1
2
2
1
1
COORDINATES
TARIFF #264
V
6471
7110
6462
6529
8476
8483
4497
5290
H
4275
4369
2634
2772
2874
2638
1385
4236
NUMBER
OF
TERMINALS
1
1
1
2
1
1
1
1
1
1
1
1
1
1
1
1
1
i
1
1
LOAD/WEEK
(CHARACTERS)
4500
5500
871250
1089000
155250
194000
17000
1475000
21250
1840000
360750
915000
450750
1140000
120000
150000
250
250
37500
46750
RATE
(CPS)
30
30
30
30
30
30
30
600
30
600
30
600
30
600
30
30
30
30
30
30
YEAR
1=1975
2=1977
1
2
1
2
1
2
1
1
2
2
1
1
2
2
1
2
1
2
1
2
-------
Table 2-2. (Cont.)
LOCATION
LEVEL IV (Cont.)
Lansing, Michigan
Duluth, Minnesota
Ely, Minnesota
Richfield, Minnesota
Roseville, Minnesota
Jefferson City, Missouri
Jackson, Mississippi
Lincoln, Nebraska
Bismarck, North Dakota
AREA
CODE
517
218
218
612
612
314
601
402
701
TIME
ZONE
E=l C=2
M=3 P=4
1
2
2
2
2
2
2
2
2
COORDINATES
TARIFF #264
V
5584
5352
5118
5781
5776
6963
8035
6823
5840
H
3081
4530
4602
4525
4498
3782
2880
4674
5736
NUMBER
OF
TERMINALS
2
2
1
1
1
1
2
2
12
12
3
3
3
1
3
1
1
1
1
1
LOAD/WEEK
(CHARACTERS)
889000
1111250
1250
1500
72750
90750
1315750
1644500
6889500
8611750
182750
228250
1497750
10600000
1870750
13250000
9000
11250
262500
328000
RATE
(CPS)
30
30
30
30
30
30
30
30
30
30
30
30
30
600
30
600
30
30
30
30
YEAR
1=1975
2=1977
1
2
1
2
1
2
1
2
1
2
1
2
1
1
2
2
1
2
1
2
-------
Table 2-2. (Cont.)
LOCATION
LEVEL IV (Cont.) :
Trenton, New Jersey
Santa Fe , New Mexico
Taos , New Mexico
Albany , New York
Buffalo, New York
Oswego, New York
Rochester, New York
Scarsflale , New York
- Cleveland, Ohio
AREA
CODE
201
505
505
518
716
315
716
914
216
TIME
ZONE
E=l C=2
M=3 P=4
1
3
3
1
1
1
1
1
1
COORDINATES
TARIFF #264
V
5164
8389
8220.
4639
5075
4759
4913
4933
5574
H
1440
5804
5786
1629
2326
2089
2195
1414
2543
NUMBER
OF
TERMINALS
1
1
1
1
1
1
1
1
1
1
1
1
1
1
2
2
1
1
2
2
LOAD/WEEK
(CHARACTERS)
37000
46250
179500
224250
513250
641500
2750
650000
3250
810000
150250
187750
13250
16500
1770250
2212750
200000
250000
883750
1104500
RATE
(CPS)
30
30
30
30
30
30
30
600
30
600
30
30
30
30
30
30
30
30
30
30
YEAR
1=1975
2=1977
1
2
1
2
1
2
1
1
2
2
1
2
1
2
1
2
1
2
1
2
-------
Table 2-2. (Cont.)
LOCATION
LEVEL IV (Cont.)
Columbus, Ohio
Oklahoma City, Oklahoma
Portland, Oregon
Chadds Ford, Pennsylvania
Harrisburg, Pennsylvania
Lewistown, Pennsylvania
Meadville, Pennsylvania
Pittsburgh, Pennsylvania
Reading, Pennsylvania
AREA
CODE
614
405
503
215
717
717
814
412
215
TIME '
ZONE
E=l C=2
M=3 P=4
1
2
4
1
1
1
1
1
1
COORDINATES
TARIFF #264
V
5972
7947
6799
5315
5363
5369
5413
5621
5258
H
2555
4373
8914
1513
1733
1869
2348
2185
1612
NUMBER
OF
TERMINALS
2
1
2
1
2
1
2
1
'8
8
1
1
2
2
1
1 .
1
1
1
1
1
1
LOAD/WEEK
(CHARACTERS)
1464500
3450000
1830500
4310000
1003250
225000
1254000
280000
4783750
5979500
521250
651500
902000
1127500
57500
71750
236250
295250
158750
198250
625250
781500
RATE
(CPS)
30
600
30
600
30
600
30
600
30
30
30
30
30
30
30
30
30
30
30
30
30
30
YEAR
1=1975
2=1977
1
1
2
2
1
1
2
2
1
2
1
2
1
2
1
2
1
2
1
2
1
2
-------
Table 2-2. (Cont.)
LOCATION
LEVEL IV (Cont.)
Columbia, South Carolina
Pierre, South Dakota
Chatanooga., Tennessee
Nashville, Tennessee
Austin, Texas
Baileys Cross Roads, Virginia
Elkins, Virginia
Huntington, Virginia
AREA
CODE
803
605
615
615
512
703
304
304
TIMli
ZONE
E=l C=2
M=3 P=4
1
2
1
2
2
1
1
1
COORDINATES
TARIFF #264
V
6901
6316
7098
7010
9005
5636
6130
5743
H
1589
5497
2366
2710
3996
1600
1452
1400
NUMBER
OF
TERMINALS
1
1
1
1
1
1
1
A
1
1
1
1
1
1
1
1
1
1
1
1
LOAD/WEEK
(CHARACTERS)
450000
37610000
562500
47010000
209500
261750
36500
45500
377000
471250
45750
57000
59500
74250
17000
21250
162000
2550000
202500
3185000
RATE
(CPS)
30
600
30
600
30
30
30
30
30
30
30
30
30
30
30
30
30
600
30
600
YEAR
1=1975
2=1977
1
1
2
2
1
2
1
2
1
2
1
2
1
2
1
2
1
1
2
2
-------
Table 2-2. (Cont.)
LOCATION
LEVEL IV (Cont.)
Richmond, Virginia
. Montpelier, Vermont
Olympia, Washington
Tacoma, Washington
Madison, Wisconsin
-
Milwaukee, Wisconsin
Charleston, West Virginia
Wheeling, West Virginia
AREA
CODE
804
802
206
206
608
414
304
304
TIME
ZONE
E=l C=2
M=3 P=4
1
1
4
4
2
2
1
1
COORDINATES
TARIFF #264
V
5906
4246
6469
6415
5887
5788'
6152
6152
H
1472
1701
8971
8906
3796
3589
2174
2174
NUMBER
OF
TERMINALS
1
1
1
1
1
1
1
1
1
1
5
1
5
1
1
1
1
1
1
1
2
2
LOAD/WEEK
(CHARACTERS)
361250
451500
308250
385250
106250
2385000
132750
2980000
190000
235000
2781750
3290000
3477000
4110000
258250
540000
322750
675000
498000
622500
60750
75750
RATE
(CPS)
30
30
30
30
30
600
30
600
600
600
30
600
30
600
30
600
30
600
30
30
30
30
YEAR
1=1975
2=1977
1
2
1
2
1
1
2
2
1
2
1
1
2
2
1
1
2
2
1
2
1
2
-------
Table 2-3 is the assumed EPA low and medium speed terminal usage distribution
as determined from the EPA raw data.
2-25
-------
Table 2-3. Assumed EPA Low Speed and Medium Speed
Terminal Usage Distribution*
Row Time
# Period
1 7:00-8:00 AM
2 8:00-9:00
3 9:00-10:00
4 10:00-11:00
5 11:00-12:00
6 12:00-1:00 PM
7 1:00-2:00
8 2:00-3:00
9 3:00-4:00
10 4:00-5:00
11 5:00-6:00
12 6:00-7:00
Distribution Over
the Week
Sun
.00033
.00066
.00165
.00498
.00498
.00171
.00498
.00498
.00330
.00165
.00066
.00012
.03
Mon
.00187
.00374
.00933
.02822
.02822
.00969
.02822
.02822
.01870
.00935
.00374
.00068
.17
Tues
.00198
.00396
.00990
.02988
.02988
.01026
.02988
.02988
.01980
.00990
.00396
.00072
.18
Wed
.00198
.00396
.00990
.02988
.02988
.01026
.02988
.02988
.01980
.00990
.00396
.00072
.18
Thurs
.00198
.00396
.00990
.02988
.02988
.01026
.02988
.02988
.01980
.00990
.00396
.00072
.18
Fri
.00209
.00418
.01045
.03154
.03154
.01083
.03154
.03154
.02090
.01045
.00418
.00076
.19
Sat
.00077
.00154
.00385
.0.1162
.01162
.00399
.01162
.01162
.00770
.00385
.00154
.00028
.07
Distribution
Within a
Single D^y
.011
.022
.055
.166
.166
.057
.166
.166
.110
.055
.022
.004
*Each Cell contains the expected or experienced connect time percentage
expressed in decimal form (i.e., 0033 = .033%) for the total time period
identified.
2-26
-------
3 . DESIGN ANALYSIS
-------
3. DESIGN ANALYSIS
This section of the report describes the design constraints placed upon the
network and the assumptions made during the study. Input data and the
major areas that have been taken under consideration are also described.
3.1 DESIGN CONSTRAINTS
A communications network as large as EPA's must be designed within certain
constraints. These define preexisting conditions and serve as the framework
in which the network must perform.
EPA plans to consolidate its Washington, D. C. based EDP resources to
provide all of its users there with a common EDP facility. EPA will also
continue to provide the computer facility at RTP. Therefore, the network
design must allow EPA users to access two geographically separate and
distinct computer facilities.
As well as servicing Washington, D. C. and RTF-based EPA users, the net-
work must also handle other EPA user locations. There are 22 of these,
none of which can be relocated to better fit the network. These 22 major
sites contain all the Level III users and include the Regional offices, the
National Environmental Research Centers (NERC's), and various EPA Labora-
tories. Refer to paragraph 2.2.2 .1.
Developing a communications network that will be sufficient to handle the
projected EPA workload for July, 1976 and also provide for immediate imple-
mentation restricts the design to current technology. This restriction
3-1
-------
greatly narrows the available design alternatives. For instance, there are now
a number of new commercial common carriers applying for tariffs from the
Federal Communications Commission (FCC) or establishing new communica-
tion links (both analog and digital) throughout the nation. It may be possible
to use some of these better established new common carriers such as MCI or
DATRAN to some extent, but it is virtually impossible to build an entire
network around them. In addition, being restricted to current technology ex-
cludes the possible use of the future packet switching networks. These may
be very cost-effective for EPA in the future, but cannot be addressed exten-
sively in this report.
3.2 ASSUMPTIONS
This section documents the three assumptions used in arriving at the
recommended network solution:
Uniformity of terminal speed
Division of workload between Washington and RTP
9 Importance of front-end unavailability
3.2.1 Uniformity of Terminal Speed
All EPA terminals were assumed to be either low speed or medium speed. The
/^~v~i
former transmit at(30/characters per second (300 baud); the latter at 600
characters per second (4800 baud).
EPA is procuring standard terminals for all remote sites. These terminals will
be capable of operating at the assumed speeds. This assumption attains the
greatest economy of scale in the assignment of ports on the front-end pro-
cessor servicing the network and attains the lowest network operating cost.
3-2
-------
Therefore EPA should standardize terminal speeds to provide the lowest costs.
If not, additional CPU ports on WATS, and perhaps even on multiplexed ser-
vice, will be required to provide the nonstandard terminals with access to the
EPA-EDP facilities. This requirement will increase the number of trunks for
WATS and multiplexed service and consequently increase the network cost.
The increased cost is greatest when a specific type of service is provided to
only a few terminals. For instance, assume there are 20 graphics terminals
operating at 300 characters per second to be serviced by WATS. Assume also
that these terminals require an additional five WATS trunks to provide 99 per-
cent availability. If the same terminals operated at 600 characters per second,
they could be serviced by the present 600 cps trunks and two additional
trunks, thus eliminating three trunks because the grade of service is related
to the load and the number of trunks via a Poisson distribution function that
assigns a greater load carrying capacity to each trunk as the number of trunks
increases. The PLANET network design tool uses the Poisson distribution
function to determine the number of WATS and multiplexor trunks required to
provide 99 percent system availability.
3.2.2 Division of Workload Between Washington and RTF
In another assumption, 60 percent of the total load upon the network was di-
rected to the Washington, D. C. computer facilities. The rest is directed to
RTP. This assumption was used in modeling the design with the split front-
end controller placements (one front-end at Washington, B.C. and one at
RTP). It was also used to determine the number of high speed channels needed
to transfer data between Washington and RTP when both front-end controllers
were at the same site. When all users were serviced by the split network,
ten percent of the load from each center was assumed to be routed to the
3-3
-------
other center to determine the number of high speed channels needed to transfer
data between the sites .
The final costs were checked carefully to determine the effect of this assump-
tion. It was found that alteration of the assumption would not change the
final recommended network design. Even complete elimination of all multiplexor
costs at both Washington and RTF, which occur because of the predicted 60/40
load split, do not reduce the overall cost of the split system below that of
locating all front-end communications controllers at a single site. Thus, the
prime effect of this assumption on the EPA network design is in determining
the number of trunks required to transfer data between Washington and RTF.
3.2.3 Importance of Front-End Resource Unavailability
Most EPA users do not require continuous access to the EDP facilities as would
an airline reservation service. Unavailability of one or two hours can nor-
mally be tolerated on an infrequent basis. Thus, ICA has assumed that re-
v
dundant equipment is unnecessary and altering this assumption significantly
increases the equipment costs. However, each network described by PLANET
has considerable fail-safe capacity. This capability exists because:
All networks require two front-end communications
controllers to handle the projected FY 77 communi-
cations load.
All networks have significant WATS capability because
the total communications load cannot be cost effectively
handled by multiplexors alone.
In all networks some remote locations serviced with
multiplexors have more than one backbone channel
3-4
-------
and have several multiplexors because of the heavy
load at the location.
These combined factors will provide a network that is unlikely to fail totally
and that will experience only mild performance degradation from communica-
tions equipment failures. (During peak usage hours, degraded performance
could become more noticeable.)
3.3 INPUT DATA
This section identifies and describes the data given to ICA by EPA and industry.
Three major types of data were received:
Network load per user location
Equipment cost parameters
A distribution of usage (load) over time
This data is input to the PLANET system and used by it in designing the
communications network.
3.3.1 Load Location Data
This data was gathered from trips to RTF and to OSI, and by EPA's workload
committee projections. The raw data was extracted from original and Work-
load Committee reports and condensed into the tables appearing in Appendix A
of this report.
Once all the raw data had been extracted and condensed, it was summarized
in a single table and used in projecting the load for each location for both
FY 75 and FY 77. See paragraph 2.2.2.3, Table 2-1. The projection was
3-5 .
-------
expressed as connect hours per month for each user location. Only locations
with usage were included in the final data summation; the others were dropped.
The summarized data (Table 2-1) was converted to input for the PLANET system
by the following procedure.
a. Each location description was used to generate the
area code, time zone, and the V and H coordinates
(in conjunction with ATT tariff #264).
b. The number of terminals was estimated by using the
equipment inventory figures, EPA's standardized
terminal projections, and average load per terminal
factors when neither of the previous two types of data
provided a projected number of terminals for the lo-
cation in question.
c. The load in characters per week was derived in all
cases directly from the projected connect hours per
month. Two factors, one for low speed terminals and
one for medium speed terminals, were developed which
were then multiplied by the connect hours per month to
arrive at the load per week figures . These factors were:
L.S. Factor = (60 min/hr)x(60 sec/min)x(30 char/sec) (4 .32 wks/mo)
L.S. Factor = (25,000 char/hr mo/wk)
M.S. Factor = (60 min/hr)x(60 sec/min)x(600 char/sec)(4.32 wks/mo)
M.S. Factor = (500,000 char/hr mo/wk)
where: L.S. = Low Speed
M.S. = Medium Speed
3-6
-------
min = minutes
sec = seconds
char = characters
hr = hours
mo = months
wk(s) = weeks
When these factors are multiplied by hours per month
data, the load expressed in characters per week re-
sults .
Operators cannot use terminals at their peak throughput
capacity. Consequently, actual connect time to trans-
mit, say, 300 characters from a 30 character-per-second
terminal is much greater than 10 seconds (the theoretical
minimum time required) and is likely to be about 30
seconds (approximately 35% transmission efficiency.)
Since the data supplied by EPA was expressed in terms of
actual or expected connect time, ICA was able to use it
directly in the PLANET modeling. Thus, the characters per
week loading figures appearing in paragraph 2.2.2.3,
Table 2-2 should not be interpreted as the number of
characters that will be transmitted, rather, they must be
modified by an efficiency factor first.
The transmission rate was specified per the assumption
described in paragraph 3.2.1.
An index figure was assigned for the data in each year. A
1 designates 1975 data and a 2 designates 1977 data. Using
index numbers this way permits the growth rate from 1975 to
1977 to vary per location, and remote cities with heavy
3-7
-------
growth have not been locked into an unrealistic
average growth figure for the overall network.
The resultant data (Table 2-2) was then entered into the PLANET system. It
is now resident in PLANET as a permanent data file and was used again and
again in additional "what if" network design analyses. Table 2-3 in para-
graph 2.2.2.3 was used as the distribution for peak loading.
3.3.2 Cost Parameters^
The design tool employed by ICA in this study, PLANET, uses several cost
parameters in performing its trade-off analysis between a WATS serviced and
a multiplexor serviced network. These parameters are:
Ml - Monthly cost for a basic remote multiplexor
M2 - Monthly cost for a multiplexor channel
M3 - Monthly cost for a basic CPU associated multiplexor
M4 - Monthly cost for a CPU associated multiplexor channel
Kl - Monthly cost for conditioning a two point line
The development of these cost parameters requires careful analysis to ensure
that the network modeling and the costs of the alternatives modeled reflect
accurately the different hardware configurations. Figure 3-1 illustrates the
hardware configuration of a network serviced by WATS. Remote terminals
would be connected to a terminal modem. There would also be a business
telephone at each remote location. This equipment is not shown in the figure.
The remote terminal operator would connect to the front-end servicing the CPU
by calling his assigned WATS number. The central site automatic answering
3-8 ....-.-
-------
CO
I
tO
Front-End
Modem Channel
Cable
Front-End
CPU
Ada
pter
Figure 3-1. WATS Network Hardware
-------
and rotary service equipment would establish a connection with a central site
modem (modem to remotes) that would be assigned to a line serviced by the
communications front-end computer. Once the link is established, data ex-
change between the terminal and the CPU can proceed.
Additional hardware would be added to the WATS network hardware in Figure 3-1
to implement a multiplexed network. This added hardware and the associated
PLANET cost parameters appear in Figure 3-2. If the modem to the remotes
were separated from the front-end modem channel cable, the added hardware
for the multiplexed network could be added directly and we would produce
Figure 3-2. Thus, the added hardware is:
the remote multiplexor, the cable to its associated high
speed modem, and the associated high speed modem
the channel cable and channel card for each channel
serviced by the remote multiplexor
the point-to-point private line connecting the remote
multiplexor to the CPU associated multiplexor
e the CPU associated multiplexor, the cable to its high
speed modem, and the CPU associated high speed modem
the channel card for each low speed channel into the
front-end computer.
The cost parameters for the PLANET modeling analysis need only reflect this
added equipment. Thus, the PLANET cost parameters are:
Ml - Monthly cost of remote multiplexor, cable to high speed
modem and remote multiplexor high speed modem.
3-10
-------
CO
1
M2
Modern T<
Remotes
5
Cable
Cha"
dare
rmel
Remote
Multiplexc
Cable
>r
Remote
Multipl
TVTori<=>m
xOr
: Ml
M4
.- ,_ . , Cable
Multiplexor
I
Front-End
Modem Channel Cable
Front-End
CPU
Lin
M3
Figure 3-2. Multiplexed Network Hardware and Associated Cost Parameters
-------
M2 - Monthly cost of the channel cable and channel card
for each channel serviced by the remote multiplexor.
M3 - Monthly cost of the CPU associated multiplexor, the
cable to its high speed modem, and the CPU associ-
ated high speed modem.
M4 - Monthly cost of the channel card for each low speed
channel into the front-end computer.
Kl - Channel conditioning costs .
The cost factors to be used in the PLANET modeling are developed in Figures 3-3,
3-4, and 3-5. In developing these costs, ICC modems and Timeplex multi-
plexors were used for the cost base because this equipment is presently in use
in the EPA network and their costs are representative of the costs of most
multiplexors and modems available today. Thus, from this analysis, the cost
parameters to be used in analyzing the asynchronous low speed requirements
are:
Ml = $142/mo (Figure 3-3)
M2 = 10/mo (Figure 3-3)
M3 = 14.2/mo (Figure 3-3)
M4 = 10/mo (Figure 3-3)
Kl = 5/mo (Figure 3-3)
Because synchronous medium speed ports cannot service low speed asynchronous
transmissions, the medium speed network requirements are analyzed separately
from the low speed requirements. Also, inclusion of medium speed loading in
the low speed WATS modeling run can substantially increase the number of
WATS ports the system may require because the medium speed terminals handle
3-12
-------
Hardware Configuration:
To
Terminals
[Modem to
Remotes
(Note 1)
X?
Asynchronous
Channel ($8)
Asynchronous
Channel Cable
($2)
TIMEPLEXT-16
($82 Includes H.S.
modem cable)
ICC
Model 24
($60)
To CPU
2400 EPS
Private Line
Service
(C-l Conditioning
$5)
This hardware configuration will support up to 8 - 300 EPS channels over the
point-to-point link to the CPU.
Basic Multiplexor Costs ($ Per Month)
$ 60 2400 bps modem
82 Multiplexor and modem cable
Ml = $142
j
Channel Costs ($ Per Month)
$ 8 Asynchronous channel
2_ Multiplexor and modem cable
M2 = $ 10
Figure 3-3. Estimated Low Speed Multiplexor Costs
3-13
-------
Note to Figure 3-3:
1. Asynchronous modems can either be obtained as part of the
multiplexor or independently as separate devices. This study was
costed on the basis of asynchronous modems being included in the
multiplexor channel costs. Data Access Arrangements would be re-
quired, however, for each channel in this case.
3-14
-------
To
~^L___
Terminals
MODEM
or
DAA
External
Device Per
Channel
Asynchronous
Channel ($8)
Asynchronous
Channel ($2)
TimeplexT-16
($82 includes cable
to High Speed Modem)
ICC Model
4800 or 7200
To CPU
$105 or $180
Private Line
Service
(C-l Conditioning)
$5
This hardware configuration will support up to 16 or 24 - 300 EPS channels
over the point-to-point link to the CPU.
Basic Multiplexor Costs ($ per month)
or
105
82
187
180
82
262
High Speed Modem
Multiplex and Modem Cable
Channel Costs ($ per month - each channel)
8
Asynchronous channel
Multiplexor channel cable
10
30 CPS Channel Capacity
4800 BPS Modem - 16
7200 BPS Modem -.24
Figure 3-4. Estimated Low Speed Multiplexor Costs
(High Density Cities Requiring Additional Capacity)
3-15
-------
Timeplex T-16 ($82)
Synchronous
Channel and Cable
($20)
Synchronous
Channel and Cable
($20)
To CPU
9600 EPS
Private Line
Service
(C-l Conditioning $5)
This hardware configuration will support two-4800 bps channels over the
point-to-point link to the CPU.
Basic Multiplexor Costs ($ per month)
$260 9600 bps modem
$ 82 multiplexor and modem cable
Ml= $342
Channel Costs ($ per month)
$18 synchronous channel
$ 2 channel cable
Ml= $20
Figure 3-5. Estimated Medium Speed Multiplexor Costs
3-16
-------
much more data than the low speed terminals.
Multiplexor costs for the medium speed data differ from those for the low
speed data. Thus, for PLANET runs with the medium speed data, the following
cost parameters will apply (Refer to figure 3-5):
Ml = $342/mo
M2 = 20/mo
M3 = 342/mo
M4 = 20/mo
Kl = 5/mo
Once PLANET runs were completed for both types of terminals, ICA combined
the results into a single network for each design alternative. The networks
thus include all costs from PLANET. The front-end costs must be developed
from the PLANET results because only those results specify accurately the
number and type of ports and, consequently, the throughput requirements for
each network front-end computer based on EPA's anticipated network loading.
3.3.3 Usage Distribution
The final input parameter required for PLANET modeling is a usage distribution
with respect to time. The relevant raw data for this distribution came from
direct interviews and fron the RTP network design report. From this raw data,
ICA determined that the EPA systems were operational seven days a week,
with Friday being the heaviest day. Greatest daily use occurred fromlO:00 a.m.
to 12:00 noon and from 1:00 p.m. to 3:00 p.m. Relative magnitudes for both
the usage over the week and over the day were developed and then multiplied
together to produce the normalized usage which would be expected for each
3-17
-------
hour of each day of the week. See paragraph 2.2.2.3, Table 2-3. This dis-
tribution is for a single site within a single time zone. The PLANET system
automatically shifts this data to reflect the time zone of the EDP center
during its analysis.
Some EPA locations operate more than twelve hours per day. This is not re-
flected in the usage distribution because it is not necessary for accurate design
analysis due to the relatively small load occurring in the extended hours.
The traffic distribution determines how many trunks or ports will be required to
service the load and what split between the full and measured trunks can be
made. The peak hour traffic and the hour by hour traffic distribution are the
significant factors that are specified in the usage distribution which permit
these determinations to be made.
3.4 SOFTWARE FUNCTIONS
This section describes the major communications software addressed in this
study. It also recommends the minimum software required for an effective EPA
communications network. .
3.4.1 Utilization Reporting
Utilization reporting is essentially an accounting function. It is typically
oriented to an environment where usage accounting is the basis for customer
billing. Some types of utilization collection and reporting are available in
most host computer software and account for CPU, channel, and device utiliza-
tion as well as connect time. Depending on the software of the host processor,
the utilization reporting can be either very detailed or simply a gross estimate.
Without proper utilization reporting, determining if the computer resources are
3-18
-------
being used efficiently becomes extremely difficult and undetermines future
planning considerations. In this study, utilization reporting was directed
towards its effect on the communications network.
During the data gathering and analysis, it quickly became apparent that EPA
utilization reporting suffered from a large number of resource providers and an
uncontrolled user environment. To alleviate this resultant lack of utilization
data, we recommend the following requirements be implemented.
a. When purchasing or leasing new equipment, or assigning
an ID, the information that should be obtained is:
1. location
2. type of terminal
3 . type of modem
4. baud rate
5 . estimated usage (both time and application)
Enter this information into the data base of the host
computer in Washington, D. C. and in RTF.
b. Place the burden of communications usage on the front-end
processors of the host computer in Washington, D. C.
Utilization accounting should include connect time and
baud rate by user and the number of characters or blocks
of data transmitted and received over communication lines.
If a concentrator is used at the RTP facility, the identical
information should be recorded there.
The recommended utilization reporting requirements would require at least a
disc drive and a magnetic tape drive attached to the front-end. The disc is
required for auxiliary on-line storage, and the tape is required for passing the
3-19
-------
utilization data to the host processor. It is also desirable to provide for on-
line utilization reporting and batch reporting via the front-end. This requires
an interactive terminal console and a medium speed printer.
3.4.2 Network Integrity and Security
Ensuring the integrity and security of an established network can be an ex-
tremely complex problem, requiring careful evaluation. A communications net-
work can be designed to provide complete data security from point of trans-
mission to point of reception. EPA does have to limit and protect the access
of certain sensitive data but only at the central processor. We recommend,
therefore, that little effort be extended to secure the communications network.
However, procedures are required to validate all users attempting to utilize
the EPA EDP facilities. Validation techniques may include User ID's, pass-
words, key words, etc., and will provide for controlled access.
Lines of sufficient quality and network monitoring are normally provided by the
common carrier. The EPA requirements do not appear to justify the expenditure
that would be required to achieve a greater network integrity.
3.4.3 Error Reporting for Networks and Terminals
A basic error reporting service is ordinarily supplied by standard communica-
tions software and augmented by common carrier services and facilities.
Additionally, the use of data conditioned lines for the major circuits considera-
bly reduces the transmission error rate. The requirements of most networks do
not warrant more sophisticated error recovery and reporting techniques than
those supplied with the basic communications services, and this appears to
be true of the EPA network. However, since the volume of traffic dictates the
use of a front-end processor and the possible use of wideband trunks between
3-20
-------
Washington, B.C. and RTF, a more complete error reporting scheme may be
realized at very little additional expense.
The communications front-end should have software mux/demux besides the
normal hardware line interfaces. This capability will allow comprehensive
software error checking when transmitting via wideband trunks or concentra-
tors. The front-end processor should also be capable of supporting an on-
line console by which a severe degradation in transmission quality may be
reported and by which lines can be taken out or put back in service via soft-
ware commands.
3.4.4 Information Routing
Since the EPA network must provide for multiple host processors, information
routing is necessary. The EPA front-end processor must be able to route
information to or from the host processor specified by each user. It is
neither practical nor economical, however, to use sophisticated routing tech-
niques such as store-and-forward or circuit switching. Rather, EPA's routing
requirements can be satisfied with a much simpler system that uses keywords
entered by the user during signon. This keyword would tell the front-end which
host processor was desired and software would then connect the user to the
specified processor, recording the disposition of that user for subsequent data
transmission and sign off.
3.4.5 Speed and Transmission Code Recognition
Dynamic recognition of both data rate and line code characteristics from a
transmitting terminal reduces the quantity of channel terminating hardware
(ports) at the computer system interface. This reduction, in turn, reduces
the likelihood that additional front-end controllers will be needed because
3-21
-------
port capacities are exceeded. The resultant reduction in hardware cost can
be significant. However, dynamic speed recognition requires that additional
consideration be given to system costs and efficiency. Use of this concept
requires that the communications control element service each port at the
highest likely data rate it must be prepared to support. This requirement
affects servicing characteristics and can limit expansion, but the actual de-
gree of constraint varies with the hardware and software. There is no doubt,
however, that dynamic, or adaptive, recognition is an asset in systems that
must react to an undefined access pattern from a number of distinctive terminals.
The RFP that defines future terminal characteristics specifies a standard data
rate of 30 characters per second and ASCII code for the low speed devices.
ICA supports this requirement as an alternative to large scale adaptive recog-
nition based upon the following practicalities.
Part of the communication system is based upon the use of time division mul-
tiplexors in which the volume of traffic is optimized for the distance to the
computer to produce an overall savings. The candidate multiplexor supports
adaptive speed and code recognition, but necessitates the installation of an
independent modem for each channel, in addition to the basic electronics pro-
viding channel control at the multiplex site. These requirements are unneces-
sary if the multiplexor can service a common data rate and line protocol.
The adaptive recognition capability required to service multiple terminal
characteristics imposes additional costs and considerations for each multi-
plexor. This cost is about $30 per month based upon standard GSA rates, and
will run approximately $2900 per month based on the projected number of mul-
tiplexors required for FY 77. In addition, added maintenance is necessary from
both the telephone company and the multiplex manufacturer.
3-22
-------
Additional space is required and all low speed terminal users must now re-
member to initiate special characters when signing onto the computer system
to differentiate data rate and other variable characteristics.
A single data rate and transmission code will simplify the hardware and soft-
ware. However, ICA recommends the initial use of adaptive recognition on
the EPA system to support those circumstances where terminal variations must
be accepted. Two specific circumstances warrant adaptive recognition.
It is very probable that despite movement by EPA towards a common terminal,
a number of IBM conversational devices will still remain in service. Also,
the use of the Tektronix graphic terminal is expected to increase. Both con-
ditions must be supported. ICA recommends that all such devices be serviced
by restricted WATS lines, and that the corresponding communication controller
ports be equipped with adaptive capabilities for speed and transmission code.
3.5 COMMERCIAL CARRIER EVALUATION
A number of commercial carriers are discussed in this section, with suitability
for use in the EPA network so denoted. None of these commercial carriers are
excluded from bidding on the recommended EPA network, but in themselves
lack the complete facilities to provide the recommended network entirely.
3.5.1 DATRAN
The Data Transmission Company (DATRAN of Vienna, Virginia, an operating
subsidiary of the Wyly Corporation of Dallas, Texas) currently offers duplex
private line data transmission on point-to-point or multipoint service at baud
rates of 2400, 4800, and 9600. Service quality of 99.95 percent of total
transmission seconds is guaranteed. DATRAN further allows their data service
3-23
-------
to be subdivided by the subscriber into multiple lower speed data channels
(multiplexing).
The tariffs for the basic DATRAN offerings include channel miles, terminating
equipment charges, local loop or channel charge, and a one-time installation
charge. A two point data service operating at 2400 baud would cost $0.75
per airline mile per month, plus $300 in recurring charges, and installation
costs of $300. The comparable 9600 baud service would cost $0.90 per mile,
$500 in other monthly recurring charges, and $400 in installation costs.
DATRAN has pending tariff applications for these same data rates on a switched
network basis at similar recurring and installation costs, but at a time-distance
rate of from $0.00015 for 2400 baud to $0.0003 for 9600 baud. In addition,
DATRAN plans wideband private line data offerings for 56 KBPS, 168 KBPS, and
1.344MBPS.
DATRAN service is attractive and competitive, but essentially unavailable to
EPA for the near future. Their operating facilities are restricted to the south-
western area of the country and it will be some time in 1975 before they can
extend service to Washington, D. C. and the major cities in the east and
midwest. Therefore DATRAN should be considered a potential supplier of both
private line and switched network data service for EPA, only after service is
available between Washington, D. C. and significant cities in its network.
Procurement of DATRAN service by EPA should result from either appropriate
figures and recommendations citing a positive cost effectiveness or other,
practical reasons. There is no present indication that DATRAN service will
include Triangle Park, N.C.
3-24
-------
3.5.2 MCI
Microwave Communications Incorporated (MCI) of Washington, D. C. was
the first specialized common carrier (SCC) to file for communications service
and the first to be granted permission to do so by the FCC. Their initial ser-
vice was between St. Louis and Chicago, and like most specialized common
carriers, the service is provided primarily through a high density microwave
system. Additional service requirements in tariffed areas beyond any micro-
wave pathway will have to be obtained from an existing common carrier.
MCI differs from DATRAN in that they provide voice transmission as well as
digital service. They are currently tariffed in 30 cities. As the pioneer SCC,
MCI has borne the brunt of the bitter litigation such concerns must face. The
efforts of both MCI and DATRAN have been retarded by legalistic refusals of
Bell operating companies to provide local interconnecting channels from the
SCC systems to a customer premises. As of May, 1974, however, the
Federal Courts have, in effect, directed the Bell System to honor all requests
for service under the SCC tariffs. . .
MCI offers their services at rates competitive to DATRAN, and both are sig-
nificantly under those of AT&T for basic private line voice grade data service.
MCI rates remain below AT&T rates even after AT&T completely revised its
private line offerings on a "Hi-Lo" basis in June of this year.
MCI has the potential to offer EPA a reliable data transmission service among
its major eastern and mid-western cities. If we assume that the SCC's will
continue to exist, a cost-capability analysis of MCI service offerings is
highly recommended since they serve that part of the nation where EPA ex-
pects some of its highest volumes of data transmission. MCI should be con-
sidered as a source of communications service for some areas of the EPA
network. ' .
3-25
-------
3.5.3 Tymshare and General Electric
Both companies offer large and sophisticated time sharing systems to users
with extensive data bases. However, it is unlikely that either could satisfy
the load that EPA would place upon it without extensive expansion of computer
facilities and communications networks . Even if either or both parties would
agree to the expansion, EPA would have to wait for the expansion to be com-
pleted and then would probably be expected to sign a long-term contract so
the expansion would be financially attractive to the time sharing company. A
long-term contract could only intensify the loss of system control and flexi-
bility that EPA should have, particularly in view of their anticipated growth of
the network. Since EPA has its own computer resources, and both the Tym-
share and the General Electric communications network are tied into their own
respective computer resources, they do not meet the EPA network requirements.
3.6 GOVERNMENT CARRIER EVALUATION
Some of the major Government carriers are described in this section and dis-
cussed regarding their suitability to the EPA network.
3.6.1 FTS
The Federal Telephone System extends throughout the country. It interconnects
civil and military agencies and provides access to the network through local
telephone connections supporting small federal offices in remote communities.
Architecturally, FTS is a modification of the CCSA (Common Control Switched
Access) private lines leased from the telephone company by many large and
physically diversified businesses.
Basically, the CCSA concept is designed for voice traffic, although some
direct link connections can provide reliable data pathways. Overall per-
3-26
-------
formance of such networks is only marginal for data transmission, particularly
when a data base must be reached by transmission through several network
switching centers. However, data transmission rates as high as 2000 bits
per second can be supported with special circuit terminations and routings.
Despite the questionable aspects of reliable performance, many government
agencies use FTS for data transmission, particularly at lower data rates. Ac-
tually, FTS is an alternative media for data transmission, since it has been
readily available to government agencies and has the extremely attractive cost
figure of 14 cents per minute of transmission. This low cost alone justifies
its wide use even when transmission reliability is marginal.
Because it is so widely used, the government is now faced with an overloaded
FTS. Voice and data traffic are contending for the same time domain of the
network, thereby degrading both forms of communication. This conflict is
not easily resolved since the demand for both voice and data traffic is rising.
There are plans to expand FTS, but this is a very involved task and will take
time to complete. Meanwhile, FTS remains overcrowded, and additional con-
nections to the system are carefully allocated, particularly in the Washington,
D. C. area.
It is therefore recommended that EPA avoid FTS as a primary access to their
data communication network, particularly on calls that would be completely
routed on FTS from terminal to data base. Further many EPA users are not,
or cannot, be serviced by FTS. Considering this unavailability as well as
versatility and reliability, WATS and remote multiplexed service remain the
most logical and expedient methods of transmitting data.
3-27
-------
3.6.2 ARPA
The Advanced Research Projects Agency (ARPA) Network is a private data com-
munications system serving the computer science community. Built and oper-
ated by Bolt, Besanek, and Newman, Inc., this type of network has been
termed a Packet Switching network. This capability is now being offered
commercially through Telenet Communications Corporation of Washington, D. C.
The first such venture by what is termed the value added carriers (VAC) was
by Packet Communications Incorporated, which provides a nationwide network
virtually identical to Telenet.
The ARPA network is designed for the exploration of new network techniques
and to interconnect and service ARPA sponsored research centers. ARPA is a
DOD sponsored activity and is not currently available for widespread use among
all the various government agencies. Although there is a great deal of pressure
being applied from a number of commercial and government concerns to make
ARPA available for expanded utilization, it remains at the present a network
that services a selective group of users. Further, since the ARPA network is
still experimental in nature, it would likely be unable to service the EPA re-
quirements. For these reasons the ARPA network is not recommended for EPA.
In this report, the overall concepts and potential of packet switching will be
addressed, since there may be future potential for EPA subscription to this
type of service from the commercial VAC's .
First, it must be understood that these concerns, like ARPA, will lease their
transmission facilities from the common carriers. Beyond that, however,
they have all developed their own methods of transferring digital data among
locations.
3-28
-------
VAC's use minicomputers as concentrators and node points to connect each
computer to at least two other computers. The interface computer is called
an Input Message Processor (IMP). The IMF's are connected in a circle, so
that information can be sent around the system, until it arrives at the
appropriate IMP serving the host computer. Data are handled in packets,
generally less than 150 bytes in size. Terminals are interfaced to the net-
work either directly to the host computer or via minicomputers known as
Terminal Input Processors (TIP). These networks are useful when a large
number of terminals must be interfaced to diverse types of host computers
and when EDP resources providing nodes are geographically dispersed.
This type of system has the advantage that the intercomputer communications
become standard while the terminals operate in their native language. The
system has two main disadvantages: the cost is generally high because of
software development for the IMP and TIP minicomputers and because of the
/
wideband lines required to yield adequate response time. Further the system
can introduce a message propogation delay of 0.5 to 1.2 second in data
transfer. Figure 3-6 shows a packet switching network.
The major advantages of packet switching include:
Multiple data transmission rates from low speed through
wide band services of 40 to 50 KBPS- All transmission is
fullduplex which allows for simultaneous transmit and
receive.
Extremely reliable error detection and correction capability
through the intelligent node principle.
Rapid, on-line access for data distribution to widely
separated locations.
3-29
-------
EPA
Seattle
IMP
TIP
TIP
I
IMP
EPA
Athens
Washington ,D.C,
Computer Center
I,
v
IMP
I
IMP
EPA
Cincinnati
-3» ft*
Triangle Park
Computer Center
To All Remote Terminals
IMP
IMP
EPA
Boston
Legend
IMP= Input message processor
TIP= Terminal message processor
Figure 3-6. Example of Packet Switching Network
3-30
-------
o Node conversion of line code, speed, and format criteria
so that terminals with unlike characteristics can commu-
nicate with each other.
0 Less constraint due to network loading and busy hour
maximums.
Reduction in the extent of physical computing and com-
munication interface and control hardware which a user
must maintain on his site.
Comparable reduction in required number of highly
qualified technical personnel to operate customer data
system.
Charges for data transfer based upon the amount of data
transferred and not upon a time/distance or facilities
mileage basis.
Example - Telenet charges under their proposed tariff of January 1974.
$1.25 per 1000 data packets transferred, each data packet
consisting of 128 characters. Forty percent for transfer at
night, and on weekends (actual hours not defined)
o Computer and terminal port charges to interface transmission
system.
0 - 9600 baud $50 per month, each
50 KBPS $100 per month, each
Port changes for dial-in access per hour
2400 baud $2.00, each
4800 baud $3.50, each
If not ARPA, certainly the packet switching concept itself appears to be worthy
3-31
-------
of serious investigation by EPA as a future medium of data transmission. With
the exception, again, of Triangle Park, most of the major cities where EPA
requires heavy data access are to be served by the VAC's. Any analysis must
carefully determine actual guarantees of the VAC as well as all costs, in-
cluding the impact of an expanded network. Certain other factors should be
kept in mind; such as, the eventual reduction of dissimilar terminals on the EPA
network, cost tradeoffs that include the anticipated number of terminals and
computer ports required by EPA, and the massive volumes of data anticipated.
The commercial VAC's must be eliminated from this study as a potential EPA
candidate network as they are not currently operational.
3.6.3 INFONET
The possible use of INFONET by EPA requires the same approach as that of
using Tymshare or General Electric. INFONET is a time sharing supplier
under contract to the government. However, some portion of the EPA require-
ment might be delegated to this system, which is provided by Computer
Sciences Corporation. INFONET is a prime candidate if supplemental service
becomes a serious consideration. As with Tymshare and G.E., however,
INFONET must be eliminated as a candidate on which the entire EPA network is
based due to the lack of EPA control and supervision over its use. Access to
INFONET is certainly adequate from most larger cities, but it is unlikely that
INFONET could serve the lesser communities in remote areas, except those
reasonably close to an INFONET network access point.
The overall requirement to provide data base access to widely scattered users
is fundamental to the EPA system. Therefore EPA must extend those communi-
cations offerings which best accommodate this reality. In addition, the data
base processing capability must be reactive to EPA's requirements. Based
3-32
-------
upon the possible size of the EPA network, it is not advisable to place this
requirement in the hands of a contractor who presently has a substantial net-
work to manage, and considering the time required to define the necessary spe-
cifics to a second party who would then implement EPA's requirements on an
existing system may exceed the cost of an EPA-installed and -operated network.
3.6.4 ADMTS
ADMTS (actually ADTS, Automated Data and Telecommunication Service) is not
a candidate for the EPA network as it is not actually an operational communi-
cations network. It is a group of professional individuals within GSA that
define, coordinate, procure for, and advise other agencies about data and
nondata communications services for individual requirements. This group
also supervises such government communication services as FTS, ARS, and a
few specialized transmission systems.
3 .7 SELECTED DESIGN TECHNIQUES
This section discusses the design techniques considered in this study and
presents the techniques ICA used in the PLANET design analysis of the EPA
network.
3.7.1 . WATS
WATS (Wide Area Telephone Service) can be described as the leasing of a spe-
cific portion of the continental switched telephone network for a specified
period of time, from an individual telephone termination. The WATS concept
divides the nation into five areas measured on a concentric basis from the
service location, but does not include service within the state. Intrastate
WATS has been calculated, where appropriate, into the network planning for
EPA.
3-33
-------
WATS is available, nationally, on a full-time measured usage time basis for
each WATS access line. Full time service provides for 240 hours per month
of usage per access line, while measured time WATS is predominantly set at
10 hours of usage per month. A very reasonable overtime charge is levied
when the usage exceeds the allowable hours in both cases. WATS allows a
user to place a large number of individual calls to any location within the
area band(s) for which he is contracted. WATS is virtually a full time service
within the 240 hour restriction. Both full and measured time offerings are paid
for at a fixed rate, per band(s), per month, regardless of the number of calls
placed within the time duration limits. WATS is a straightforward and versatile
means of providing for random communication service over a wide area, and is
expected to be the most significant service used to configure the network since
it closely fits EPA's requirements.
WATS can be arranged as IN WATS or OUT WATS from any point, such as a
computer location, so that calls can be originated or received at that location
without incurring toll charges for each call. There is no difference in cost
or use guidelines between these two categories. Recent changes to the WATS
tariff has added a restriction that all calls on any access line must be at
least one minute long or a surcharge, similar to the overtime charge, is
applied.
Because EPA has many sites with low usage, the PLANET system uses WATS
service to provide cost effective access to the EPA-EDP facilities for low
usage sites. The WATS service calculated for the EPA network uses the best
mixture of Full Time, Measured Time and Intrastate WATS, based on the usage
statistics of the EPA terminals included in this study. WATS overflow onto
the toll network is also included when necessary to extend terminal access
to a computer site when an additional WATS termination is unjustified.
3-34
-------
WATS is fully supported by the Telephone Company and usage trends can be
obtained regularly so that practical and economic control can be maintained
to allow adaptation to changing usage .
3.7.2 Multiplexors, Concentrators, and Front-Ends
Multiplexors, concentrators, and front-ends are the predominant types of
hardware used in communications networks.
3.7.2.1 Multiplexors
Multiplexors are channelizing devices attached to a private line telephone
circuit connecting several locations . They electronically derive additional
transmission channels from the voiceband spectrum of the circuit to which
they are attached by using either frequency division or time division. Mul-
tiplexors can be applied to both switched network and private line applications.
Multiplexors have one function: to reduce communication line costs by com-
bining data from several terminals for transmission over a single communications
link. They thus, eliminate the requirement fora separate DDD, WATS, or
private line from each terminal to the data base. It must be remembered that
while multiplexors reduce the number of circuits between the remote terminals
and the processor, they do not reduce the number of interface ports. Another
fact to bear in mind is that multiplexors have no storage or buffering capability.
Some time division products feature a multiple character storage to allow for
timing realities in system environments, but they do not contain memory as
such. They are merely hardware devices that enhance the communications en-
vironment by providing additional pathways for real time data throughput. The
absence of memory makes them inexpensive for the overall capabilities they
provide to the user.
3-35
-------
Frequency division is normally used with data rates of 10 to 30 cps or when
several multiplexors are combined in a multipoint configuration. Time division
is normally used with higher data rates of 120 to 600 + cps. It provides more
channels per circuit, but is generally more expensive than frequency division.
Present state-of-the-art time division multiplexors sometimes can be used as
lower level concentrators at considerably less cost than a true concentrator.
The PLANET system used time division multiplexors in the EPA network ana-
lysis because they can transmit a higher rate of data than frequency division
multiplexors. Almost without exception, the EPA remote locations have very
high data volumes. Therefore, the lower cost per data bit caused by passing
more data on more channels at higher data rates on a single pathway offsets
the increased cost of the multiplexor hardware. This increased cost is further
offset by the additional reliability, flexibility, and control features in this
more sophisticated hardware.
The time division principle employs both character-interleaving and bit-inter-
leaving schemes for passing digital data. While there is some delay in
character assembly for either approach, there is negligible delay in frequency
division units, but again, this scheme has a practical transmission speed and
channel limitation. Generally, character interleaved multiplexors are less
transparent than bit interleaved for channel intermixing of baud rates and line
cost disciplines. Transparency in this case refers to the ability of the mul-
tiplexor to pass the data between two terminal points without being aware of
the existence of that data. This fact is important where the originating terminal
must dial into the distant computer via the multiplexor and proceed through
the hardware-software handshaking routines necessary to establish a connec-
tion. Frequency division multiplexors are transparent to both code and data
rate, but the most recent time division products have effectively overcome
their previous limitations in this area. Excellent time division and frequency
division multiplexors are available-. Both have their advantages of application
3-36
-------
but the latest time division units are rapidly overcoming several factors which
they once conceded to frequency division techniques.
There is no difference in effect on individual channels by either product on the
host computer interface. However, time division units allow the processor or
its communications controller to use software for demultiplexing, thereby re-
ducing equipment costs and presenting a single data channel with the total
transmission spectrum.
Figure 3-7 shows a typical time division point-to-point multiplexor configuration.
3.7.2.2 Concentrators
Concentrators, like multiplexors, derive lower line costs . They service a
number of terminals by funneling data from them into high speed trunks that
link the output side of the concentrator to the front-end of the computer at a
distant site. A concentrator not only reduces individual terminal line lengths,
hence cost to the ultimate computer site, but it also reduces the number of
input ports required at the computer site. The computer system communica-
tions controller must have the software capability to demultiplex the incoming
data into identified blocks that can be recognized as the transmission from
the individual remote terminals.
Most concentrators today are computers of varying capability. They provide
some communications control and in-transit storage, thereby proportionally
reducing the need for similar software in the main computer system. Concen-
trators can store and forward information, change formats, translate line
codes, edit data on a dynamic basis, control input from terminals, and si-
multaneously produce journal (log) with a corresponding traffic analysis
based on a prespecified set of parameters. Where major hardware or terminal
3-37
-------
CITY "Z1
CITY "Y"
o
B
A
E
D
Front-End
Communications
Controller
Of ' ''
Main
Processor
System
co
CO
oo
A - Time Division Multiplexor
B - Multiplexor modem which provides clocking, data rate, and determines the .
capacity of the multiplexor
C - High speed, private line trunk circuit linking the multiplexor system to the main
processor system and replacing the multiple lower speed circuits feeding the
multiplexor from various terminals
D - Multiplexor output channels in individual ports of front-end communications
controller
E - Multiplexor input channels from remote terminals. These channels can accom-
modate a variation of built-in and external modems depending upon channel
speed and complexity.
F - Switched network and private line circuits feeding into multiplexor from terminals
G - External modem
Figure 3-7. Typical TDM Equipment Configuration
-------
subsystems change frequently, the programmable concentrator allows con-
tinuation of previous operation by emulating the characteristics of the new
or previously installed devices, obviating replacement of central system
software.
Few devices are available specifically designed as concentrators, but many
manufacturers provide minicomputers modified for this use, and with impressive
capability. They can be arranged for very straightforward concentration duties
requiring a minimum of software and memory requirements. Depending upon
traffic, grade of service, and total system considerations, concentrator
capability can be enhanced step-by-step to where they can perform many con-
trol tasks and relieve the data base system of almost all communications su-
pervision. This capability reduces software requirements at the host system
but increases software and hardware sophistication and therefore cost at the
concentrator. Concentrators are usually maintained and supported by the vendor.
Figure 3-8 shows a communications link with a concentrator.
The application of concentrators within the EPA network cannot be determined
at this time. A good deal of exacting statistics are required to determine the
deployment of concentrators within a network. More detailed information in
regards to EPA's actual terminal usage, data volume statistics, and processor
requirements is needed. To attempt to place concentrators in the network
configuration instead of multiplexors, based upon the information presently
available, could lead to unwarranted time and expense as concentrators are
basically minicomputers requiring extensive software, hardware, maintenance,
and manpower support. Although the use of concentrators may be very viable
for the EPA network, it is recommended that they not be included in the initial
installation of the proposed network. However, once the proposed network is
in operation, utilization statistics should be gathered and evaluated to .
3-39 ; " '-'
-------
CITY "X"
CITY "Y"
GJ
i
CONCENTRATOR
0
O
oo
O
FRONT-END
COMMUNICATIONS
CONTROLLER
OF
MAIN
PROCESSOR
SYSTEM
A - Concentrator ports servicing X number of private line and switched network
communication circuits connecting the remote terminals of the system to the
concentrator.
B - Concentrator disc file for intransit storage, data block buffering, and control/
utility program residency.
C - High speed, private line trunk circuits linking concentrator to main processor
system, and replacing the many individual, lower speed circuits feeding the
concentrator from various terminals .
D - Switched network fall back link to allow continuous but degraded operation in
cases of unreliable operation or failure of primary trunk circuits.
Figure 3-8. Concentrator Network Equipment Configuration
-------
determine the existence of any heavily used data pathways that justify the in-
stallation and use of a concentrator.
3.7.2.3 Front-End Controllers
A wide variety of capability exists in this area with performance closely tied
to cost. Some front-ends are not processors such as the IBM 2703-series,
but are essentially peripheral devices that provide data signal recognition,
code conversion, and some character and sequence recognition. They are not
an independent entity. But they can be used satisfactorily depending upon the
types of terminals serviced, the communications channel interfaced, and the
transfer demand placed upon them by the terminals and the system application.
Their use should be discounted, particularly with respect to the apparent EPA
requirement and the total capability of the host processor.
The most effective means of relieving the host processor of communications
system support and control, however, is a true programmable minicomputer
front-end dedicated to this task. An effective programmable front-end.places
the least demands upon the main processor system software and manipulates
and controls the I/O data flow with minimal interrupt servicing in the main
processor. The recommended EPA network has a definite requirement for a pro-
grammable front-end device because of the throughput capacity requirement as
well as the return information flow for interactive access. Throughput efficiency,
capacity, and cost are the major criteria in programmable front-end evaluation,
along with a number of other primary functional considerations.
The ability to interface multiple computer systems
simultaneously with balanced but not degraded I/O
servicing.
3-41
-------
The number of collective ports, or communication line
terminations, that the front-end can service, for both
medium speed and low speed transmission rates.
The flexibility and complexity of the port hardware
itself, and the range of communication interfaces it
can support.
The number and types of peripheral devices that can be
connected to the front-end, such as disc, tape, high
speed printer, etc.
The line codes, transmission protocols, and individual
features of various terminal systems that can be supported
by existing software, as well as the adaptability of the
control software for modification.
The major product emulation and plug-to-plug capability
of the front-end so that its addition to existing mainframes
does not require major software changes, and so that it
enhances the processing availability of the total system.
e The extent of utility and diagnostic software available
with the product, including such communication oriented
programs as traffic analysis, terminal servicing statistics,
error recognition, and transmission channel quality
monitoring.
The total buffering capacity of the front-end and the level
to which it can independently carry on communication ser-
vicing functions during temporary main processor unavail-
ability. This consideration is tied directly to the levels
and time intervals of graceful degradation or fail soft
characteristics.
3-42
-------
Figure 3-9 shows a front-end controller and its functions.
The EPA network requires programmable front-end processors to handle the
large number of ports and to effect the necessary information routing. They
will be described in detail in the Phase II specifications.
3.7.3 Leased Lines
Leased, or private, lines are communication circuits which are routed between
or among specific locations, normally in use by a single user. These facilities
are paid for at a fixed rate per month depending upon their total circuit mileage,
and the complexity of the communications hardware devices which the pro-
viding common carrier must install with them to meet the performance criteria
of the users application.
The primary advantage to this type of communications facility is their availa-
bility on a 24-hour day, 7-day a week basis, in a dedicated environment, at
a fixed price.
Leased, or private, lines are excellent pathways for data communications.
Their characteristics allow much versatility to support critical information
transfer. Leased lines are an easily controlled, high quality transmission
medium. For the most part, leased lines are used where priorities and urgency
require unobstructed availability between the locations they serve. Further,
they are economical to use when much data must be transferred. Recent tariff
changes have changed the lease line pricing structure for commercial users,
but the Federal Government still has this service to almost everywhere in the
country at amazingly low cost. Leased lines are priced on a monthly basis
at established rates per circuit mile, and can be obtained both on an inter-
state and intrastate basis. .
3-43
-------
CO
I
D
o
3=>
A - Front-end communications controller
B
C
D
Host processor system in same physical area
Full duplex, memory channel interface connecting both units
E -
Communications controller ports interfacing various types of
modem hardware in turn serving communication channels from/to:
1 - Switched Network
2 - Private Line
3 - Time Division Multiplex System
Peripheral devices used by front-end system if it is a true, inde-
pendent processor. These devices would not be valid if front-end
was merely a pre-processing peripheral of the host processor
system.
Figure 3-9. Front-End Communications Controller Equipment Configuration
-------
Leased lines are contemplated for two areas in the EPA network. They will
provide the connecting, or backbone, circuit for remote multiplexors and will
link certain medium speed, heavy usage terminals to their supporting computer
data base. Where feasible, they will also be used to provide the computer-to-
computer data transmission links. Leased lines are employed in this manner
because they allow unrestricted data transfer without contending for or absorbing
switched network facilities, and because they are not constrained to the lesser
data rates which limit the average switched network connection.
The initial concept for the EPA network will use leased lines as point-to-point
pathways between the front-end controllers and the time division multiplexors.
The use of multipoint leased lines linking more than two locations could be
more extensively studied if the requirements become evident.
3.7.4 Foreign Exchange (FX) Service
FX service combines switched network and private lines. It is always configured
as a two point facility. When used in a data communications environment, it
is installed so that one end of the private line terminates in a computer or
communications controller port. The other end terminates in a central exchange
office serving the general area where the distant terminals are located. With
foreign exchange service, as with WATS, multiple users can access a direct
pathway to a data base incurring minimum or no toll calls from the access
termination of the foreign exchange line. The use of FX is restricted to where
several terminals could use a foreign exchange service without serious.con-
tention and at less cost than a WATS access line. Since almost all EPA lo-
cations would require multiplexing equipment to avoid serious contention among
users., FX service is not used as a design technique in the EPA network.
3-45
-------
3 .8 THE EPA PLANET SYSTEM
A PLANET System has been developed for EPA. This system is a set of programs
from the PLANET library that have been tailored and integrated for EPA's spe-
cific needs. Direct access to the system is available to ICA via the time
sharing service on which the library is maintained. The tailoring and integra-
tion to the precise needs of EPA has several advantages:
PLANET is tailored with all of the programs required to
work with EPA's design techniques . And if new design
techniques are desired, they can be readily added.
o PLANET is tailored for direct control of all pertinent
system variables, allowing extensive sensitivity
analysis, or what if engineering.
PLANET provides just the information needed.. This out-
put tailoring enables selective printout of results,
skipping extensive reports when they are not needed, but
producing them when they are .
PLANET is tailored to work with the data on traffic, CPU
throughput, and terminal locations that is available now,
but structured to accept a more detailed data base should
it become available later.
PLANET is tailored for automatic file transfer of data from
program to program. This greatly simplifies and speeds
up extensive sequential analyses.
PLANET avoids the dangers of canned programs that, to
cover many system types, are too general for any one
in particular.
3-46
-------
o PLANET allows for new tariffs or program changes that
may occur to the designer.
Based on the considerations identified in the preceeding subsections, ICA
has structured PLANET to provide careful evaluation of two network design
techniques. They are:
o Providing users access to EPA's;EDP re source-providing
centers via WATS service.
Providing users access to EPA's EDP re source-providing
centers via a leased multiplexed network.
The final network design combines these alternatives relying exclusively on
one or the other. They are evaluated with the WATSYN and MUXSYN program
components.
WATSYN - This PLANET program component takes the EPA loading and load dis-
tribution data and synthesizes a WATS system to service the network. Traffic
falling within free call local areas is segregated from the total traffic so it
does not influence either interstate or intrastate WATS band configurations.
WATSYN calculations also optimize full and measured WATS service. The
primary function of WATSYN is to decide in which WATS band EPA should have
its trunks.
MUXSYN - This PLANET program component uses the same input as WATSYN in
locating multiplexors to reduce the costs of the WATS services. Time division
multiplexors are used in the EPA calculations. .
The interaction of these PLANET program components in the analysis process
is shown by Figure 3-10. The master input file containing the locations of
3-47
-------
t
I
Master
Input
File
1
C Start J
1
.L >
^
\
^/
i
Net It
Logical
Editor
..
ff
)
/
__[.__
L i
I
X
Design
Parameter
File
j
1
« 4 .
-s
5
Working
Q Terminal
Locations
o Network L<
/ WAT3YN or \
^r
1
1
/ i
*a
f
; i>
WATSYN
1
S<* ,., .
'
1
i_
i Yp S
X^^MUXE
x^.
S
!YN ./
X
*\
«*>
I
1
1
1
-h
-4
._L_
\i \
/Da
rt»W
\ A W
V
^^&ll ^^"
k«_
/ ^
j,
? Jy
MUXSYN
1
i
_ _j
\
I
\
i
i
1
I
1
i
i I
Planet
Data Base>
foWats Tab/e Files
o Hi-Lo Tahriffs
Poison Blockinc
( End J
I
Analysis
Results
.J
Figure 3-10. Interaction of PLANET Program Components and Data Base
3-48
-------
the EPA terminals and the data loads on each terminal are input to the NETIT
program component. NETIT allows logical editing of the master input file
data. Because EPA is anticipating that users will use either the EDP facili-
ties in Washington, Research Triangle Park (RTP) or both, with a split in the
loading to be about 60 percent for the Washington facilities and 40 percent for
the Research Triangle Park Facility, NETIT permits the loading of each remote
location to be proportionally distributed to either EDP facility.
Once the master input file has been processed by NETIT and a working file has
been produced, the WATSYN or MUXSYN may be run with the data in the working
file. The first run with WATSYN produces Bands 1 through 5 and the intrastate
band. The breakdown is:
The bands to be equipped with WATS trunks
Costs for full time access lines
Costs for measured time access lines
Measured time overtime costs
The network designer makes the next run with MUXSYN. This run places a
multiplexor and updates the working file, removing the load now handled by
the multiplexor. Now another run may be made with WATSYN or MUXSYN using
the updated working file which has the decreased load. However, MUXSYN
is usually run again and again until no more multiplexors can be placed. Once
this process is complete, the designer computes the overall network costs
(the WATS costs and the multiplexor costs). The final data represents a net-
work design using WATS and multiplexed service techniques most effectively.
The PLANET model is reviewed by ICA design personnel to assure conformance
with EPA's network design objectives. This step will probably result in
3-49
-------
altering the PLANET results because even though they represent the lowest
cost network design, they may not necessarily be the best solution for EPA.
Thus, overriding factors and other practical considerations that affect network
design and functions and that cannot be incorporated into program logic, have
been accounted for in the EPA network planning.
3 .9 DESIGN ALTERNATIVES I, II AND III
Three network design alternatives appear to be the most promising candidates
for EPA implementation. In ICA's proposal to EPA, several network configuration
alternatives were identified. Further analysis resulted in three candidate net-
work configurations for PLANET modeling and final cost comparison. Each
configuration is a direct variation of the centralized computer control concept
presented in ICA's proposal. The predominant factor in determining the network
structure is the type of node participation. A network node can be a provider
of EDP resources, a user of EDP resources, or a combination of both.
Centralized structures are very appealing when a relatively large number of
sites need to reach only a few sites which participate as resource providers.
On the other hand, distributed structures are most appropriate when all par-
ticipate as both EDP resource providers and EDP resource users. Two EPA
facilities function as the major resource providers to a large number of remote
sites. Thus, the centralized structure best fits EPA's needs.
A centralized structure would have a network communications controller (NCC).
This communications control computer would act as a focal point for monitoring
network activity, access to all EDP resources, and controlling network as
well as EDP facility usage. The NCC could be interfaced directly to the EDP
processing computers or it could be interfaced to them via high speed data
links. If it were interfaced directly, it would act as a communications front -
3-50
-------
end for the applications processing computers. If the NCC were interfaced
via communications links, it could emulate either a remote concentrator or a
high speed remote terminal. In the network configurations analyzed, the NCC
may perform both the functions of a front-end and a remote concentrator.
The three network configurations proposed for EPA are shown by Figure 3-11.
In alternative no. 1, the NCC is in the same facility as the Washington
Computer Center (WCC). The NCC would be connected to the RTF computers
by high speed data channels and interfaced directly to the WCC computers.
Thus, it would function as a front-end for the WCC computers and be a remote
concentrator or high speed remote terminal to the RTF equipment. The users
would be interfaced to the NCC via a WATS multiplexed network determined
most cost effective by PLANET. In this alternative, 40 percent of the total
network load would be routed to RTF via high speed data channels.
In the second alternative the NCC is with the RTF computers but it is otherwise
like alternative no. 1. In this alternative 60 percent of the front-end load is
routed to Washington via high speed data channels .
The third alternative uses one NCC at RTF and one at WCC. Under this con-
figuration 60 percent of the workload is routed directly to Washington, B.C.
and 40 percent of the workload is routed directly to RTF.
3.10 ALTERNATIVE ANALYSIS
The design alternatives were analyzed using the PLANET output. The first
step was to determine the effects of front-end processors (network commu-
nications controllers) on network costs. Thus, for each design alternative,
the PLANET results describing the number of low and medium speed ports re-
quired at each EDP facility were extended as shown in Table 3-1. A breakdown
3-51
-------
Alternative #1
,
wcc
RTF
NCC
USERS
Alternative #2
USERS
Alternative #3
USERS
Figure 3-11. EPA Alternative Network Configurations
3-52
-------
Table 3-1.' Front-End Loading and Data Transfer Estimates
CO
I
en
oo
ALTERNATIVES
1 - Washington
2 - RTF
3 - Washington
RTF
Combination
Low Speed Ports
Local
89
27
57
14
71
Other
256
319
181
150
331
Total
345
346
238
164
402
Medium Speed Ports
Local
19
12
14
7
21
Other
66
76
52
37
89
Total
85
88
66
44
110
Est. Peak Char.
Throughput
L.S.
3623
3633
2499
1722
4221
M.S.
33150
34320
25740
17160
42900
% Of
. Load
Transferred
40
60
10
10
-
Total Peak
Char. Load
Transferred
Per Second
14709
22772
2824
1889
-
# Channels
To Transfer
Data - M.S.
14
22
3
2
5
-------
of ports servicing local data loads and ports servicing the remote data load is
provided in the table for both low speed and medium speed ports. The number
of low speed and medium speed ports .required to service the load by front-end
controllers located in either Washington or in RTF is virtually identical.
With the split system, the number of ports for both low speed and medium
speed access to the systems is slightly higher than that required for either
alternative 1 or 2. This is an indication that the split system is not as effi-
cient in servicing the total EPA network load as a single combined system at
either Washinton or RTF. However, the size of the EPA FY 77 data load is
such that each alternative will require two front-end processors, with the data
load split between them. These processors would have equivalent costs be-
cause the throughput requirements on each would be essentially the same and
the basic equipment required thus would be the same for each alternative.
In the next step, the PLANET costs are compared for each design alternative
using WATS only and using a WATS-MUX combination. The results of this
analysis are given in Table 3-2 . The WATS-MUX network requires more front-
end ports than the WATS-only network, so the costs of these added ports
include incremental front-end hardware costs (see Table 3-3) which must be
included in the WATS-MUX costs for accurate comparison. Thus, the front-
end hardware costs for the remote channels were multiplied by the additional
number of ports required to develop the increased front-end port costs appearing
in Table 3-2. The totals in this table demonstrate that the relative costs of a
WATS-MUX network are lower for each alternative than the relative costs of a
WATS-only network. The savings realized in every case is substantial.
As a final step, the WATS-MUX combination network for each alternative was
compared. In this final analysis, the PLANET cost for the WATS-MUX com-
bined network was augmented by the cost of the channels needed to transfer
data between Washington and RTF, and the costs of modems and front-end
3-54
-------
Table 3-2. WATS Only/WATS-MUX Relative Cost Comparison
ALTERNATIVES
1 - Washington
2 - RTF
3 - Washington
RTF
Combination
WATS Only
L.S.
170,417
215,579
109,063
97,205.2
206,268.2
M.S.
89,195
94,443.4
57.154.3
42,918.9
100,073.2
Total
259,612
310,022.4
166,217.3
140,124.1
306,341.4
WATS-MUX Combination
L.S.
73,874.8
82,438.4
57,410.7
51,837.1
109,247.8
M.S.
58,647.3
63,285.3
47,079.1
34,487.7
81,566.8
Sub-Total
132.522.1
145,723.7
104,489.8
86,324.8
190,814.6
Increased
Front-End
Port Costs
2,533.76
2,817.92
2,036.48
1,468.16
3,504.64
Total
$135,055.86
$148,541.62
$106,526.28
$ 87,792.96
$194,319.24
CO
en
en
-------
Table 3-3. EPA Front-End Estimated Port Hardware Costs
Multiplexor - Handles 8 Line Adapter
Multiplexor Units
$75/mo.
$1.18/mo./channel
Line Adapter Multiplexor - Handles 8
FDX Channels
$116/mo. $ 14.5 0/mo./channel
Channel Interface - Handles 1 Full
Duplex Channel
$8.00/mo ./channel
Total Cost Per Channel
$23.68/mo.
3-56
-------
port hardware. (See Table 3-4.) The grand total cost thus developed includes
all network costs except the costs attributable to the basic front-end con-
troller hardware and software. However, we have previously determined that
these basic costs would be equivalent. These network costs are summarized
in Table 3-5. This cost summary clearly demonstrates that alternative no. 1
is the most cost effective solution to the EPA network problem, and is there-
fore recommended. It has both front-end processors (network communications
controllers) located in Washington, D. C. and all EPA users are serviced
from that central site. Forty percent of the data received by the network front-
end processors is routed to RTP. The recommended network is described in
greater detail in the next section and its ability to cost effectively replace the
present EPA network is determined.
3-57
-------
Table 3-4. Incremented Hardware Costs
Transfer Channel Costs/Synchronous - 1200 CPS (9600 Baud)
Multiplexor - Handles 8 Lines Adapters
Line Adapter Module - Handles 2 Full
Duplex Channels
Computer Hardware Cost Per Channel
Modem Cost for Transfer Channels (Two
1200 CPS Modems)
Mileage Costs
Total Transfer Channel Cost
$75/mo.
$63/mo.
$9.375/channel
$31.5/channel
$40.875
$500/channel
$446.50/channel
$987.375/channel
Remote Channel Modem and Front-End Hardware Costs
300 Baud Low Speed Modems (30 CPS-103A)
Front-End Hardware
Total Low Speed Channel Costs
4800 Baud Medium Speed Modems (600 CPS-208A)
Front-End Hardware
Total Medium Speed Channel Costs
$25/channel
$23.68/channel
$48.68/channel
$125/channel
$40.875/channel
$160.875/channel
NOTE: All costs are expressed in terms of dollars per month.
3-58
-------
Table 3-5. Alternative Cost Summation
WATS - MUX Combination Network
Alternatives
I Washington
II RTF
III Washington -
RTF
Combination
PLANET Costs
Low Speed
$73,874.80
$82,438.40
$109,247.80
Medium Speed
$58,647.3
$63,285.30
$81,566.80
Costs of Channels
to Transfer Data
$13,823.25
$21,722.25
$ 4,936.88
Modem & Front-End
Port Hardware Costs
L.S. M.S.
$16,794.6
$16,843.28
$19,569.36
$13,674.38
$14,157
$17,696.36
Grand
Total Cost
$176,814.33
$198,446.23
$233,017.09
3-59
-------
4. RECOMMENDED NETWORK - ALTERNATIVE #4
-------
4. RECOMMENDED NETWORK - ALTERNATIVE #4
This section presents the analysis and supportive information determining the
cost-effectiveness and suitability of Alternative #4 as the selected EPA network
and describes the FY75 and FY77 configurations. The information is presented in
a series of tables (Table 4-1 through 4-11) which provides a detailed description
of the EPA communications network, its interrelationship and costs. The informa-
tion contained in these tables are for both the FY75 and FY77 configurations and
include the following:
Cost Summary
Loading and data transfer estimates
o Required local telephone service
WATS service and costs
Time Division multiplexor system requirements
- Number and speed of point-to-point lines
- Number and speed of multiplexors
- Number of low and medium speed ports required
by city
- Extra port capacity by city
Port assignments
- Washington, D .C .
- RTP, N.C.
4-1
-------
Table 4-1. Ccnrounications Network: FY75/FY77 Cost Sumnary
WATS - MUX Coribination Network
YEAR
FY75
FY77
PLANET Costs
L.S.
$56,858.80
$70,372.80
M.S.
$39,986.25
$54,985.80
Sub-Total
$96,844.05
$125,358.60
Cost of
Transfer
Channels
$8,886.38
$12,835.88
Network
Modem
Costs
$15,825
$22,575
Reconfigura-
tion
Savings
($7,307)
($15,075.30)
Grand
Total Cost
$114,248.43
$145,694.18
i
N)
NOTE: Communications Costs NOT included in ICA Study:
1. Modems at remote terminals
2 . Local telephone company charges for:
a. Business lines associated with WATS service
b. Data Access Arrangements for use with low speed channel modems or multiplex
system at remote sites and for low speed channel modems at CPU site where
direct cable connection is not feasible.
c. Rotary selector equipment at CPU site for IN-WATS call distribution across
communications front-end ports.
d. Special assembly equipment or modification where required by local policies.
3. Installation,, disconnect, and move charges associated with #2 , above.
4. Multiplex vendor charges associated with initial service and installation of
their products.
-------
Table 4 - 2. Comrunications Network: FY75-FY77, NCC loading and Data Transfer Estimates
LOCATION
. W
c
c
»,
R
T.
P
YEAR
FY75
FY77
FY75
FY77
Low Speed Ports
Local
66
89
60
100
Other
162
229
~
Total
228
318
60
100
Medium Speed Ports
local
15
19
11
18
Other
43
60
^
Total
58
79
11
18
Est. Peak Char
Throughput
L.S.
2394
3339
630
1050
M.S.
22,620
30,810
4290
7020
Total Peak
Char. Load
Transferred
Per Second
10,006
13,660
_ *
_ *
#
Channels To
Transfer
Data-M.S.
9
13
^
CO
*Transfer load included in VJCC figures listed above.
-------
Table 4-3. Local Telephone Service for EPA Coirputer
Site at Washington, B.C. and RTF
LOCATION
W
C
c
R
T
P
YEAR
FY75
FY77
FY75
FY77
PORTS
Low Speed
66
89
60
100
Medium Speed
15
19
11
18
4-4
-------
Table 4-4. FY75 Composite Listing Cornmunications
Services for EPA Network
IN-W&TS to EPA Computer Site
Washington, D.C.
(Low Speed)
Full Time Access Lines
240 Hours Per Month
Measured Time Access
Lines
10 Hours Per Month
Band 1
Band 2
Band 3
Band 4
Band 5
Intra-State
4
2
6
2
(Medium Speed)
Full Time Access Lines
240 Hours Per Month
Measured Time Access
Lines
10 Hours Per Month
Band 1
Band 2
Band 3
Band 4
Band 5
Intra-State
3
0
5
4
4-5
-------
Table 4-5. FY77 Composite Listing Communications
Services for EPA Network
IN-WATS to EPA Computer Site
Washington, D.C.
(Low Speed)
Full Time Access Lines
240 Hours Per Month
Measured Time Access
Lines
10 Hours Per Month
Band 1
Band 2
Band 3
Band 4
Band 5
Intra-State
6
2
(Medium Speed)
Full Time Access Lines
240 Hours Per Month
Measured Time Access
Lines
10 Hours Per Month
Band 1
Band 2
Band 3
Band 4
Band 5
Intra-State
1
4
4
7
4-6
-------
Table 4-6. WATS Summary for FY75
(Low Speed Terminals)
Grade of Service .01
Band
1
2
5
6
7
$-Full
Trunks
575.00
0.00
8475.00
0.00
990.00
$-Full
OA
0.00
0.00
0.00
0.00
0.00
$-Meas
Trunks
580.00
370.00
1740.00
310.00
0.00
$-Meas
OA
593.18
0.00
3970.92
0.00
0.00
Grand Total $17604.10
(Medium Speed Terminals)
Grade of Service .01
Band
3
5
6
7
$-Full
Trunks
2600.00
5085.00
0.00
225.00
$-Full
OA
0.00
0.00
0.00
0.00
$-Meas
Trunks
1540.00
1450.00
620.00
0.00
$-Meas
OA
3746.50
3036.21
169.22
0.00
Grand Total $18471.90 |
4-7
-------
Table 4-7. WATS Cost Sumnary for FY77
(Low Speed Terminals)
' Grade of Service .01
Band
1
3
5
6
7
$-Full
Trunks
1150.00
0.00
5085.00
0.00
1335.00
$-Full
OA
0.00
0.00
0.00
0.00
0.00
$-Meas
Trunks
580.00
1540.00
1740000
310.00
0.00
$-Meas
OA
458.37
2973.46
3926.27
0.00
0.00
Grand Total $19098.10
(Medium Speed Terminals)
Grade of Service .01
Band
2
5
6
7
$-Full
Trunks
860.00
6780.00
415.00
285.00
$-Full
OA
0.00
0.00
0.00
0.00
$-Meas
Trunks
740.00
2030.00
465.00
0.00
$-Meas
OA
931.06
4396.96
44.50
0.00
Grand Total $16947.50 . |
4-8
-------
Table 4-8. FY75 Composite Listing Ccanmunications
Services for EPA Network
Time Division Multiplex System
All Channels and Circuits Terminating in EPA Computer Site
Washington, D.C.
Site
Seattle
Denver
Chicago
Atlanta
San Francisco
Portland, Ore.
Corvallis, Ore.
Las Vegas
Cincinnati
Athens, Ga.
Grosse He, Mich.
Richfield, Minn.
New York City
Boston
Namagansett, R.I.
Rochester, N.Y.
Columbus, Ohio
Roseville, Minn.
Dallas
Philadelphia
Madison, Wise.
Gulf Breeze, Fla.
Jackson, Miss.
Kansas City, Mo.
Full Duplex
Point-to-Point
Private Lines
2400
EPS
2
-
1
1
1
1
1
1
1
1
1
1
1
1
-
1
' 1
1
4800
EPS
-
-
-
1
-
-
-
-
1
-
-
-
-
-
-
_
-
1
-
_
T
7200
EPS
-
-
-
1
-
-
-
-
-
-
-
-
-
-
-
-
-
-
9600
EPS
2
-
2
1
-
-
1
1
-
1
-
1
1
-
-
-
1
1
-
-
"
Low Speed
Channels
MUX
2
2
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
PORTS
9
9
11
12
6
6
6
4
15
7
5
5
4
4
4
7
4
11
4
4
5
Medium Speed
Channels
MUX
2
-
2
1
-
-
1
1
-
1
-
1
1
-
-
-
1
1
-
-
PORTS
3
-
3
2
-
-
2
2
-
2
-
2
2
-
-
-
2
2
-
'
4-9
-------
Table 4-9. FY77 Composite Listing Coranunications
Services for EPA Network
Time Division Multiplex System
All Channels and Circuits Terminating in EPA Computer Site
Washington, D.C.
Site
Seattle
Denver
Chicago
Atlanta
San Francisco
Portland, Ore.
Corvallis, Ore.
Las Vegas
Cincinnati
Athens, Ga.
Grosse lie, Mich.
Richfield, Minn.
New York City
Boston
Namagansett, R.I.
Rochester, N.Y.
Columbus, Ohio
Poseville, Minn.
Dallas
Philadelphia
Madison, Wise.
Gulf Breeze, Fla.
Jackson, Miss.
Kansas City, Mo.
Full Duplex
Point-to-Point
Private Lines
2400
EPS
_
-
-
-
1
1
1
1
-
-
-
1
1
1
1
1
1
1
-
-
1
1
1
1
4800
EPS
_
1
2
-
-
-
-
-
-
1
1
-
-
-
-
-
-
-
1
- .
-
-
7200
EPS
-
-
1
-
-
-
-
1
-
-
-
-
-
-
-
-
-
-
-
-
-
-
9600 .
EPS
4
1
4
2
.-
-
1
-
1
1
2
-
1
1
-
-
-
-
2
1
-
-
-
"
Low Speed
Channels
MUX
-
1
2
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
-
1
1
1
1
1
PORTS
15
10
19
17
7
6
6
6
17
13
13
4
7
5
5
4
4
8
6
14
5
4
4
6
Medium Speed
Channels
MUX
4
1
4
2
-
-
1
-
1
1
2
-
1
1
-
-
- '
-
2
1
-
-
-
PORTS
7
2
8
4
-
- -
2
-
2
2
4
-
2
2
-
-
-
-
3
2
-
'
4-10
-------
Table 4-10. EPA Cities With Multiplexor Port Expansion Capacity
FY75
City
Seattle
Chicago
Cincinnati
Atlanta
Philadelphia
Number of Spare Low Speed Channels
6
4
1
4
4
TOTAL
19
FY77
City
Denver
Chicago
Atlanta
Cincinnati
Athens, Ga.
Grosse He, ,Mich
Philadelphia
Number of Spare Low Speed Channels
6
13
7
7
3
3
2
TOTAL
41
4-11
-------
Table 4-11. Network Cormunications Controller Port Assignnents
pOCATION
W
C
C
R
T
P
'.
YEAR
FY75
FY77
FY75
FY77
*
Total Ports
L.S.
228
318
60
100
M.S.
58
79
11
18
Local Ports
L.S.
66
89
60
100
M.S.
15
19
11
18
WATS Ports
L.S.
20
24
-
-
M.S.
21
20
-
-
Multiplexor Ports
L.S.
142
205
-
-
M.S.
22
40
.
*Does not include ports to transfer data between Washington and RTF.
These ports operate at 9600 baud and there are 9 in IY75 and 13 in FY77.
------- |