Post on 07-Jan-2016
description
Can ISPs be Profitable Without Violating Network Neutrality?
Amogh DhamdhereConstantine Dovrolis
Georgia Tech
Amogh DhamdhereNetEcon 2008
Disclaimer
This is not a game theory talk
Amogh DhamdhereNetEcon 2008
The Network Neutrality Debate
Recent Trend: Large amounts of video and peer-to-peer traffic on the Internet
Access Providers (AP) deliver content to users Recent trend: Not profitable Flat rates, commoditization of Internet access
Content providers (CP) generate the content Profitable (think Google)
Tension between AP and CPs: “Network neutrality” debate
Traffic shaping/prioritization by ISPs
Amogh DhamdhereNetEcon 2008
A Technical View
Previous work Mostly non-technical Emotional debates in the press, painting APs as villians
But what about the underlying problem: Non-profitability of Access Providers?
Our approach: A quantitative look at AP profitability Investigate reasons for non-profitability Evaluate strategies for the AP to increase profit
Amogh DhamdhereNetEcon 2008
Modeling AP Profitability
Three AS types: AP, CP and transit provider (TP)
Focus on the AP AS links
customer-provider (customer pays provider)
peering (no payments) AP and CPs can transfer
traffic either through customer-provider or peering links
TP TP
APAP
CPCPCP CP
CP
CP
CP
CP
Amogh DhamdhereNetEcon 2008
Baseline model AP and CP connect to the TP as customers N users of AP, charged a flat rate R ($/month)
Flat rate prices decrease due to competition
Transit pricing: 95th percentile of traffic volume 95th / mean = 2:1 for normal traffic, 4:1 for video1
More video means higher transit payment by AP
AP users: Heavy tailed distribution of content downloaded per month
High variability in AP costs
1Norton’06: Internet Video: The Next Wave of Massive Disruption to the U.S. Peering Ecosystem
Amogh DhamdhereNetEcon 2008
AP Strategies – Charging Charging strategies
AP charges “heavy hitters” according to volume downloaded
AP caps heavy hitters AP charges CP (non-network
neutral) Charging strategies are
disruptive AP users may depart,
depending on existing competition
Parameter d determines shape of departure probability curve
AP cannot control customer departure probability
Amogh DhamdhereNetEcon 2008
AP Strategy – Charging Heavy Hitters
Threshold T to identify heavy downloaders
Charge “by volume” for heavy hitters
c(D) = D*R/T, where download amount D, threshold T, flat rate R
Customer departure probability depends on T and d
AP’s profit is sensitive to customer departure probability
For some values of d, no threshold gives larger profit than baseline !!
Amogh DhamdhereNetEcon 2008
AP Strategies - Connection
Connection Strategies AP caches content from CPs AP peers selectively with CPs
Goal: Save transit costs paid to the transit provider Does not increase the AP’s revenue
Non-disruptive AP does not risk losing customers
Amogh DhamdhereNetEcon 2008
AP Strategy – Cache CP Content AP caches content from
some CPs locally Saves transit costs, as
content is served locally Increases local costs
incurred by the AP Critical parameters:
Fraction of content that can be cached (h) and cost incurred for caching (s) Live content cannot be
cached! Profit is sensitive to h, s
Amogh DhamdhereNetEcon 2008
AP Strategies – Peering with CPs
Peering selectively with Content Providers can save transit costs, without risk of losing its users
But, peering is not free Fixed, traffic dependent costs
Peering cost classes for CPs “Low”: CPs at the same geographical location/IXP “Medium”: CPs at nearby location/IXP “Hard”: CPs in different continents
Cost benefit analysis: r = Estimated benefit/Estimated Cost Peer if r > R
Amogh DhamdhereNetEcon 2008
AP Strategies – Peering with CPs
Optimal point exists for the cost-benefit threshold R AP controls the factor R
Significant reduction in AP costs with selective peering
Greater benefit with fewer CPs (more traffic from the largest CPs)
AP can leverage expansion by large CPs
Amogh DhamdhereNetEcon 2008
Conclusions
Network Neutrality research should also focus on the underlying problem: Non-profitability of ISPs
How can ISPs be profitable in spite of increasing traffic, heavy-hitter users and video traffic ?
Charging schemes that target heavy hitters may not work in the presence of competition in the AP market Profit highly sensitive to customer departure probability Out of the AP’s control
Connection strategies such as peering selectively with Content Providers seem promising Completely under the AP’s control
Amogh DhamdhereNetEcon 2008
Thank You !
Amogh DhamdhereNetEcon 2008
AP Strategy – Capping Heavy Hitters
AP caps download rate for users
Limits the total amount of traffic handled
Saves transit costs and local costs, but does not increase revenue
AP profit depends on customer departure probability
For some values of d, no threshold gives larger profit than baseline !!
No significant improvement over baseline
Amogh DhamdhereNetEcon 2008
AP Strategy – Charging CPs
AP directly charges the top sources of content (CPs)
Increases revenue, but violates “network neutrality”
Subject to customer departure due to discriminatory practices
AP profit depends on customer departure probability