TCOM 541 Session 6. Network Design Process Define Problem Collect data Refine data Choose/build/ run...
-
Upload
dustin-gervase-bryant -
Category
Documents
-
view
215 -
download
0
Transcript of TCOM 541 Session 6. Network Design Process Define Problem Collect data Refine data Choose/build/ run...
TCOM 541
Session 6
Network Design Process
DefineProblem
Collectdata
Refinedata
Choose/build/run Tool
Examineresults
Each stage influences bothnext and the previous stages
Define Problem
• What are we trying to do– Build a network from scratch?
– Replace an existing network
– Expand an existing network?
– Merge two networks?
– Solve a reliability/performance problem?
– Implement a new technology?
– Reduce costs?
– Etc., etc
Collect Data
• What types of data are available?
• How complete is the data?
• How reliable is the data?
• Very often, discoveries will be made at this stage that will lead to a redefinition of the problem
Refine Data
• Check to identify and eliminate bad data– E.g., circuits that go nowhere, circuits that are
suspiciously cheap or expensive, ...
• Fill in gaps– Estimation, modeling, or guesswork ...
• Put into form required by model
Choose/Build/Run Tool
• Choice of design tool is heavily influenced by type of problem and data availability
• Only the simplest problems are solvable without use of automated Tools
• The user is to a large extent at the mercy of the algorithms built into the Tool
Examine Results
• ALWAYS examine the design outputs carefully– “Sanity checks”
– Global statistics (hops, latency, cost distribution, ...)
– Look for anomalies
– Vary design parameters
• Repeat process until satisfied ... or time/budget runs out
Acquisition vs. Design
• At present, users are often more concerned with acquiring services (e.g., VPN) than designing a network as such
• This presents a completely different set of problems and opportunities– Essentially, the vendors do the design, the user
picks the best
Acquisition vs. Design (2)
• But data is still essential– Accurate information to vendors– Accurate projection of costs to support
evaluation
• Managing the competitive process is critical– Private users have much more freedom than
Government – Federal Acquisition Regulations are restrictive
Acquisition Case Study
• Back in the 1970’s, the General Services Administration ran what was essentially a private phone company, called Federal Telephone Service (FTS)– Long distance voice only– On-net only– Leased lines, switches, …– Not very efficient
• Cost $0.27 to $0.40/minute, depending on how the accounting was done
• About 30% to 100% more than commercial rates
FTS2000
• FTS was replaced in 1988 by FTS2000– Services-oriented 10 year contract with
significant volume banding (I.e., lower prices at higher volumes)
– Mandatory use by agencies– Voice, packet-switched data, dedicated circuits– Two vendors, mandated 60/40 split– Two internal recompetitions at years 4 and 7
FTS2000 – Scope
FTS2000 Network A
FTS2000 Network B
0 375 million minutes/month0 $44.0 million/month
0 1.7 million users0 4,200 locations
Customer Premises
Equipment(GSA)
Local Exchange
Carrier
Inter-Exchange
Carrier
Local Exchange
Carrier
Customer Premises
Equipment(Agency)
User User
Service Delivery Points
1989 1990 1991 1992 1993 1994
30
25
20
15
10
5
0
Negotiated Price Decreases
FTS
FTS2000 Cutover Complete
Price Redetermination
Switched Voice Price
(Cents Per
Minute)
FTS2000 Cutover Begins
1995
Year End
FTS2000 Price History
Current FTS2000 price is 1.5 cents/minute less than best
commercial price
FTS2000 Replacement
• Strategy development started in 1994 for planned award in 1998
• Called FTS2001
• Situation changed– No mandatory use– Technology advances– Deregulation/local competition
FTS Program• Objectives
– Provide high quality telecommunications services that meet users’ needs
– Leverage the large volumes of Government traffic to obtain the best prices
• Characteristics– Flexibility as a means to deal with uncertainty
(technology, market, regulatory, requirements)– Maximize competition– Agency choice– Market oriented– Rely on private sector– Use commercial best practices
Strategic Alternatives
Continue Current Compre-hensive
Contracts (Alternative 1)
Integration Contractor
(Alternative 2)
Span-Specific Contracts
(Alternative 3)
Regional Compre-hensive
Contracts (Alternative 4)
Integrated Business Process
Solutions (Alternative 5)
Service-Specific
Contracts (Alternative 6)
Service/Span-Specific
Contracts (Alternative 7)
Government-wide Approach?
Individual Agency
Acquisitions (Alternative 8)
Agencies Agree on Coordinated Approach
Partitioned
by Serviceby Span
Comprehensive
by Span and
Service
Yes No
Strategy Choice
• High-level agency (customer) working group
• Decision support tools (Analytic Hierarchy Process)
• Inputs from oversight bodies
Strategy Choice (2)
• Comprehensive contracts– 1 or 2 contracts, 8 years– Expanded technology suite– Internal recompetition– Provision for local access competition– Need for significant commitment
• $750M Minimum Revenue Guarantees
• Agency commitment to support
Pricing Structure - Objectives
• Pricing structure must support objectives:– Award two or three contracts with almost
equal prices over likely range of traffic (e.g., 20% to 60%)
– Don’t leave money on the table• Want total FTS2001 cost close to lowest possible
– Facilitate internal competition
Pricing Structure – Problems
• Whatever range of traffic volumes is used for evaluation, offerors will likely bid at least these three price break points:
– Just below the lower limit– Just below the upper limit– Just below 100%
• Probably will bid more breaks
Pricing Structure – Problems (2)
• Two or three awards probably will cost more than single award, in the short-term
– Second and third vendors for any service will likely raise prices
• Declining price-volume curves
– FTS2000 experience - initial 2-vendor award cost ~15% more over first 4 years
• Break-even not achieved until Year 7
Pricing Structure Problems (3)
• Effect worse when 100% prices are significantly lower than 20% - 60% prices
• Possible exception if vendors have significantly different prices for high-volume services– Effective candidates limited to SVS, 800/900
and DTS
Pricing Structure - Example of Possible Bids
EvaluationRange
100%
Price
Volume
Offeror A
Offeror B
50%
Pricing Structure - Example of Resulting Problems
EvaluationRange
100%
Price
Volume
Offeror A
Offeror B
50%
Two herecost more than one here
Premium paid for twocontractors
A is better within theevaluation range, but B is better at 100%
Pricing Structure - The 100% Problem
• If 100% is not evaluated, GSA leaves an unknown amount of money on the table
– And may inadvertently award 100% to a higher bidder– Someone will work out how much and publicize it
• If 100% is evaluated, offerors will most likely structure their bids to drive to a winner-take-all outcome
– Cannot recoup excess costs by later internal competition
– May actually result in significant savings to Government over first 4 years
Pricing Structure - Summary
• As initially constrained with price-volume banding, GSA could not win
Pricing Structure – Partial Solution
• Change constraints:– Flatten prices by eliminating or severely
restricting volume discounts• Reduces gaming by offerors
• Eliminates government volume-band chasing after award
• Conforms more closely to commercial practice
• Simplifies pricing and billing
Still A Problem
• Elimination of volume bands removes the 100% problem
• But does not move closer to ensuring two providers with low and nearly-equal prices
Complete Solution
• Move to two-stage award process• Initial award for nominal (but not guaranteed)
50% of network• Publish prices at aggregate level• Allow offerors to bid new prices for remaining
nominal 50%– Structure this as a contract modification for winner of
round one– Means these prices apply to his initial 50% whether or
not he wins the additional 50%
30
RedetermineCompRange
EvaluateResubmission
EvaluateProposals
EstablishCompRange Negotiations
andResubmissio
n
Request BAFO
EvaluateBAFO
Repeat BAFOBAFO Received Make
Award(s)
Publish Prices
StopRequest APOIf Only One
Award
Initial Evaluation BAFO Evaluation APO Evaluation
ReceiveProposals
EvaluateAPO
Repeat APOAPO Received
Make Award
Stop
AWARD PROCESS
Stop
Stop
Results
• Sprint won Round One with very low prices– Chose not to bid for second 50%
• WorldCom won second 50% with prices a few percent lower than Sprint
• Objectives achieved.
FTS Switched Voice Service Unit Prices
0
10
20
30
Oct
-88
Oct
-90
Oct
-92
Oct
-94
Oct
-96
Oct
-98
Oct
-00
Oct
-02
Cen
ts p
er M
inut
e
FTS Switched Voice Service Prices (log scale)
1
10
100O
ct-8
8
Apr
-89
Oct
-89
Apr
-90
Oct
-90
Apr
-91
Oct
-91
Apr
-92
Oct
-92
Apr
-93
Oct
-93
Apr
-94
Oct
-94
Apr
-95
Oct
-95
Apr
-96
Oct
-96
Apr
-97
Oct
-97
Apr
-98
Oct
-98
Apr
-99
Oct
-99
Apr
-00
Oct
-00
Apr
-01
Oct
-01
Apr
-02
Year
Cen
ts p
er M
inut
e
2
3
4
5
20
30
40
50