1、Interconnection: An Economic Perspective,Peyman Faratin (CSAIL) Steven Bauer (CSAIL) David Clark (CSAIL) Bill Lehr (CSAIL) Arthur W Berger (Akamai,CSAIL) Patrick Gilmore (Akamai) Tom Wilkening (Economics),Interconnection Problem,AT&T - Carter phone & Hush-a-Phone (blocking) . 2002: Madison-River - V
2、onage (blocking) 2005: Cogent-Level 3 (disconnecting) 2006: AT&T - Google (tiering) 2007: T-Mobile2 (blocking) ICE (Farrell and Weiser), Agency (Milgrom et.al), Entry Story - because of lack of quality competition in interconnection Two-Sided Markets (New Institutionalist Model) A model of value-flo
3、ws - demand information Market failures “middlebox”/overlays entry Interconnection discrimination incentives (given cost-allocation mechanism),Industrial Organization: Two-Sided Markets,Generative: Design aidBusiness ModelDescriptive: future regulatory thinking,Causal Hypothesis of Interconnection P
4、roblems,Architecture,IO & Contracts,Information & Behaviors,Outcomes,The Trinity: Institution, Strategies and Outcomes,Institutionarchitecturecontractpolicy,OutcomesScalability, Resilience, ConvergenceFairness, Innovation, Profitability,Strategic Agents,Transfer Distribution Ambiguities (“we know ho
5、w to route packets but not money”),AS1,AS2,content,$,AS1,AS2,$,Ambiguities Galore,AS1,AS2,content,$,AS2,$,AS1,content,AS2,$,AS1,Solution: Bi-lateral Volume-Based Contracts,Retail market (bursty): Flat-rate Peak-rate tiered pricing Wholesale market (better aggregation “deeper in”): Full transit Trans
6、fer level = non-linear Transfer structure = asymmetric Peering Transfer level = 0 Transfer structure = N/A Emerging mechanisms: Paid-peering & Partial Transit Distribution of Fixed and Usage pricing,Architecture,IO & Contracts,MIT,http:/,End-Hosts Bear Cost of Transport,$,$,$,$,$=0,$=0,No E2E Accoun
7、ting for Tastes,Coordination Failures Has Led to E2E Market-Failures,Market-Failure Induced CDN Entry AKAM: 20,000 servers,900 networks,70 countries,750 cities, serving 15% of content,Strategies and Outcomes,Contracts,Information & Behaviors,Outcomes,The Trinity: Institution, Strategies and Outcomes
8、,Institutionarchitecturecontractpolicy,OutcomesFairnessGrowthProfitability,Strategic Agents,Who Should Pay Who? Primitive = Value-Flows,ISP,i,j,Pi,Pj,pi,pj,I,II,III,IV,(0,0),“Free Goods”,Q: what is the optimal price structure for ISP to maximize profits?,eyeball,Content Provider,Value-Flow Discrimin
9、ation,Q: what is the optimal price structure? A: Depends on:Relative size of value flows (cross-market externalities)Fixed / Per transaction pricesSingle v.s Multi-homing,pi,pj,45o,ISP,i,j,Pi,Pj,eyeball,Content Provider,Established commercial web-server $ ISP $ eyeballs,Complementarities/Interaction
10、s: Multi-Product Markets,Value-Flows/Externalities: Chicken-Egg Problems,Two-Sided Markets,But platform has to solve “chicken-egg” Problem: if there were more women, then more men would come, more women would come, more men would come,. discrimination is welfare enhancing. “ladies nights”,Non-Discri
11、mination Institution,Strategic Agents,Does Institution Implement Desired Outcome?,Rule (motivated by “fairness”): No bars can access discriminate based on sex Q: Does rule implement a “fair” & innovative outcome in the presence of strategic actors?A: No. Institution is “fair” but gives no growth inc
12、entives. Neutrality rule is not neutral with respect to growth tussle between objectives,Result of Rule: Closes Some Markets, Others Grow but Inefficiently,Strategic Preferences of Content Providers & Users,ISP,$,Eyeballs,Growing WebServer,ISP,$,Eyeballs,Established WebServer,$,ISP,?$,Eyeballs,Publi
13、c WebServer,?$,$,Strategic Agent Preferences: The Platform (in Presence of Externalities),Platform (ISP/CDN) solves for efficient prices: market price level ( ) and price structure Profit maximizing pricing structure in presence of externalities is often discriminatory (subsidize one side of the mar
14、ket to stimulate demand on other side - c.f. bar) Strong incentives to discriminate,Network Neutrality Law or Current Architecture & Protocols,Institution“the architecture cant / shouldnt do that”“no price discrimination for same service”,OutcomesFairnessGrowth,Strategic Agents (1:Customer, 2:Conten
15、t Provider),3: Platform: ISP,Unintended Outcome of Institution: Market Closures,Externalities Create Surplus Expansion Opportunities (v.s. Capture),Traditional (one-sided) Price discrimination Discrimination increases the profits of the monopolist but may open some markets that would otherwise be cl
16、osed. platform intermediaries in a TSM seek to maximize profit by transferring surplus from seller to consumer thereby growing the market Growth on one side of the market induces growth on the other, creating surplus that can be captured,Market-Failure Induced CDN Entry: Akamai: 20,000 servers, 900
17、networks, 70 countries, 750 cities, serving 15% of content,Architectural Tools We Provide,The real question is how to architect for it: Change in demand in i market / change in demand in j market Source-destination discrimination App discrimination Per packet/per flow bit discriminate Encryption . T
18、here is a delicate tradeoff involved in how much information we provide and how much we lose/gain in objectives we are interested in,Architecture,IO & Contracts,Information & Behaviors,Outcomes,Conclusion,Interconnection Not only a L2, L3 problem Contract engineering and value-flows Agents use mecha
19、nisms strategically Tussle over outcomes Open Questions: Preferences over outcomes/objectives CDN Tipping and Market-Power 2 tiered Internet? Externality Information for monitoring and regulation Industrial Organization A tool for architecture & policy,Future: ICWG,Data War Stories/cases Peering of
20、video Exclusivity contracts Games being played . Quantities and prices data to support theory data to build theory Informative process to all Designers ISPs Policy makers,Peymanmit.edu,Auxiliary Slides (I),Information and Strategic Games,Competition: Peering+Transit Strategic Interactions,All compet
21、e to: establish and maintain peering Competition over: Eyeball Networks Content Colo CP (Apple iTunes, Microsoft,) Stub ASs (Yahoo, Google,) Non-stub Tier2 content (transit providers to content Stub AS),“Normal” Business Strategy of LE-LC,Strongest Peering Incentives,Assume LE-LC interconnect under
22、peering LCs problem is to keep ratios,LE-LC Strategies,Observations: Eyeballs are fixed, content can move (switching costs of content is lower) perception of bargaining power by LE LE doesnt care about being out of balance & in fact wants to be out of ratios so it can demand payments (paid-peering),
23、“Equilibrium” in Establishing New Peering between Strategic Networks,LE-LC Peering Establishing Strategies,LE strategy: LC asks to peer (or upgrade peering facilities to keep abreast of traffic flows) LE refuses and demands higher settlements (paid-peering) because:it is LC who is out of ratios and
24、causing costs Operational costs (AOL) Precedence settings leads to economic loss on the long-run Most LCs refuse to pay, but some do concede. Some content owners on LC who doesnt concede switch to LCs that do.,LE-LC Strategies: Vertical + Horizontal,LCs Counter strategy (“chicken”): If LE refuses to
25、 peer/upgrade peering then LC sends some traffic via transit Punishing strategy: LC bears P2 (which may even be above cost of P1), but LE has to pay P3 Condition: Strategy only works if both LC&LE are transit customers of tier1. If LE has peering with tier1 & LC sent via transit then LC would in fac
26、t be helping LE because LE would look bigger to tier 1,P1,P2,P3,LCs Strategy to Keep Ratios: Sell Low-cost Transit (Poaching: Vertical+Horizontal),LCs strategy: Peering link is full-duplex and LC is mostly outbound To keep ratios LC needs to pull sell transit to SE Poaching SEs by setting P2 at or e
27、ven below cost LE P2P traffic to SE goes via LC,P2,LCs Strategy is Reactive and Proactive,P2,T,Ratio Balancing Needs Create Poaching Competition, Downward Pressure on Transit Prices and Quality,Margins of gain of poaching strategy to maintain peering shrinks as P2 falls Excess reductions of P2 lower
28、s quality/performance of transit because incentives of LC to manage are eroding?,P2,Salient Economic Features,Dynamic efficiency (innovation) Operator IO is highly complex (no clear upstream/downstream) Behavioral: Direct & indirect network Effects Unobservability Coordination failures,Auxiliary Sli
29、des (II),TSM Model,How ISP Determines its Optimal Price Structure: Geometry of the Problem,ISP,i,j,Pi,Pj,pi,pj,I,II,III,IV,(0,0),“Free Goods”,Q: what is the optimal price structure for ISP to maximize profits?,eyeball,Content Provider,Value-Flow Discrimination,Q: what is the optimal price structure?
30、 A: Depends on:Relative size of value flows (cross-market externalities)Fixed / Per transaction pricesSingle v.s Multi-homing,pi,pj,45o,ISP,i,j,Pi,Pj,eyeball,Content Provider,Established commercial web-server $ ISP $ eyeballs,Total Consumption,is “native” demand,demand of i due to demands of j,Total
31、 Consumption,network externality term (how much purchases in j market affects purchases in the i market),Benchmark: eji = eij = 0,pi(pj),pi,pj,pj(pi),1/2,1/2,Po = (1/2,1/2),pj,eji=0,pi,pj,eji=3/4,pi,pj,eji=11/10,pi(pj),pj(pi),Architectural Guide,eij a potential candidate for value-flow proxy,Value-F
32、low and Structural,Q: what is the optimal price structure? A: Depends on:Relative size of cross-group externalitiesFixed / Per transaction pricesSingle v.s Multi-homing,pi,pj,45o,ISP1,i,j,Pi,Pj,usr,Google,ISP2,usr,Assumptions,Networks tariff: Charges to i market for subscription Charges to j market
33、for traffic termination i market single-homed Makes single either-or decision competition between platforms for i market i chooses network that maximizes its surplus j market multi-homed Makes independent join decisions no competition between platforms for j market j puts more weight on network bene
34、fits of being in contact with widest population of i market than transaction costs of multiple platforms,Equilibrium Tariff (M. Armstrong),Low subscription charges to i market and high termination charges to j market Equilibrium termination charges to j market maximizes i market and networks profits
35、 and ignores j market welfare.,pi,pj,I,II,III,IV,Multi-homing Reduces Competition and Welfare,Single-homing side is treated well, m-homing sides interest are ignored at equilibrium (i is even cross-subsidized) “Competitive bottleneck”: even if market for content users is highly competitive, so that
36、profits of networks are lowered, there is no competition for providing services to content providers.,Engineers Provide Tools to Firms: Design-Evaluate Cycle,IO methodology: puts economics (back) into the design consideration, but after protocol designAllows “comparative statics” - “what happens to
37、welfare if we change the institution”Build testable models to ask “what-if” questions on efficiency-fairness tradeoff,Future,Competition for ideas and incentives Strategic agents will use technical & regulatory tools to their economic advantage FIND (2006): 3/10 economic (CABO, Virtualization, Archi
38、tecture of all fiber networks) Highly recommend talking to economists & regulators SIGCOMM 08 Workshop? MITs Interconnection Working Group David Clark, Steven Bauer, Bill Lehr, Peyman Faratin, Akamai,Markets,ISP,i,j,Pi,Pj,qj,qi,usr,Google,Geometry of the Price Discrimination Problem,pi,pj,i has rela
39、tively more externality impact on j,j has relatively more externality impact on i,MC,Demand,Each market has a continum of consumers willing to buy one discrete unit of good (transport service) Let v be arbitrary willingness to pay of an individual Then D(p) is the market demand,CUs Market Demand,vp
40、dDdpi,- Vi,- Qi,D(pi),- v,Maximum market size (in absence of network externalities),Maximum service value (in absence of network externalities),Total Consumption,is “native” demand,demand of i due to demands of j,Total Consumption,network externality term (how much purchases in j market affects purc
41、hases in the i market),Measures,Spill-over/TS network externality = cross-price (i to j) contribution to sales in j market.,Measures,Importance of the spill-over effects,Externality of CPs to CUs,As CPs use more transport then CUs max. service value for transport increases,- Vi,- Qi,D(pi),Externalit
42、y of CPs to CUs,- Vi,- Qi,CU value increase,Consumer Surplus,- Vi,- Qi,- Vi,- Qi,Si = QiVi / 2,Sji = (eji Qj)Vi /2 = QiVi / 2,Surplus Division v.s. Capture,Third-Degree Price discrimination Firms offer nonlinear prices to mixed markets force heterogeneous consumers to self select (Peak-rate pricing?
43、) Mechanism differentially extract consumer surplus and transfer it to the seller platform intermediaries in a TSM seek to profit by transferring surplus from seller to consumer Growth on one side of the market induces growth on the other, creating surplus that can be captured,Monopoly Pricing in Ab
44、sence of Network Externality (Po)(monopoly sets prices in the two markets independently, implicitly assuming eij = eji = 0),Monopoly CUs Profits,- Vi,- Qi,- Qi / 2,- Vi / 2,Monopoly Pricing with Network Externality (P*)(monopoly sets prices in the two markets interdependently, eij eji 0),Assume:j ma
45、rket (CPs) demand for transport is inelastic i market (CUs) demand for transport is elastic eji eij 0 qi / pj 0 (Positive TS, spillover, effect),qj=eijDi(pi),qi=Di(pi),pi,qi=ejiDj(pj),qj=Dj(pj),pj,+,+,+,+,+,-,-,-,eji eij 0, i j,qj=eijDi(pi),qi=Di(pi),pi,qi=ejiDj(pj),qj=Dj(pj),pj,+,+,+,+,+,-,-,-,pi,qi,qj,pj,qi,pi,pj,qi,pi,Asymmetricity in Externalities,Now vary the relative influence of CP CU eji eij 0,Benchmark: eji = eij = 0,pi(pj),pi,pj,pj(pi),1/2,1/2,Po = (1/2,1/2),pi(pj),pi,pj,pj(pi),1/2,1/2,I,II,III,IV,eij = 1/3,pj,eji=0,pi,pj,eji=3/4,pi,pj,eji=11/10,pi(pj),pj(pi),
