A Modeling Framework to Understand the Tussle between ISPs and Peer-to-Peer File Sharing UsersMichele Garetto - unitoDaniel R. Figueiredo - EPFLRossano Gaeta - unitoMatteo Sereno - unitoIFIP Performance 2007

P2P is starting to dominate Internet TrafficP2P represented 60% of InternetTraffic at the end of 2004

CacheLogic Advanced Solutions for P2P Networks

Presentation | WCW 2005

Impact on Service ProvidersP2P is the dominant protocolIn excess of 92% of P2P traffic crosses transit/peering linksP2P protocols will aggressively consume any available bandwidth capacityDue to P2Ps symmetrical nature on average 80% of upstream capacity is consumed by P2PP2P affects QoS levels for ALL subscribersService Providers can not afford to block or restrict P2P ISPs must intelligently manage P2P - blocking and shaping doesnt workP2P is driving consumer broadband uptakeand broadband is driving P2P uptake

The ISP perspective vs P2P:threat or opportunity ?P2P traffic: friend or foe ?friend: driving force for adoption of broadband access by the usersfoe: overwhelming amount of traffic What is the best strategy to manage P2P traffic in my network ?Try to kill it ?Do nothing ?Educate it ? How ?

Strategies to manage P2P trafficAcquire more bandwidthBlock P2P trafficTraffic shaping (e.g., priority to non-P2P)Pricing schemes based on user traffic volumes / bandwidth capsNetwork caching / customized P2P application within ISPApplication-layer redirection of P2P traffic

Our contributionSimple model to analyze the impact of P2P traffic on an ISP networkBasic insights about system dynamicsPerformance evaluation tool to evaluate different strategies to manage P2P traffic and guide ISP choices

Network scenarioUser issues a query Solved outside consumes bandwidth BdSolved inside does not use bandwidth Bd Exploitation of traffic locality

System modelobject retrieval probability:

User Utility Function We express the utility of user i as: -1-0.5 0 0.5 1 0 0.2 0.4 0.6 0.8 1User utilityprobability of successful object retrievalshape parametersubscription cost

ISP Utility Function We express the utility of the ISP as: fixed chargecost per unit of external bandwidthprofit from subscribers feeThe ISP starts the service only if

Modeling traffic localityProbability that there exist at least one internal copy of an object replicated r times in the system Number of internal copiesNumber of external copiesProbability to download from internal (existing) copy (i.e., exploit traffic locality) :

Modeling object replicationWe need to compute the number r of replicas of an object at the time the object is requested, taking into account:We have developed an analytical technique based on Poisson shot noise processes

Different popularity of contentsTemporal evolution of object popularity Introduction of new contentsCancellation of replicas by the users

Modeling object replicationExample: (2 objects) popularitytimeA video from the newsA popular songt1t2request

ResultsAssumption: n identical users N = 50 millions request rate by user (object/day) introduction of new contents by user (object/day)

Minimum required external bandwidth:

The impact of object replication (r) 0 1000 2000 3000 4000 5000 0 10000 20000 30000 40000Minimumrequired bandwidth Bmin(objects/day)Number of users, nr = 500r = 1000r = 1500

The impact of efficacy in exploiting traffic locality ( ) 0 20000 40000 60000 80000 100000 0 100000 300000 500000 700000= 0.25= 0.5= 0.75= 1.0Number of users, nMinimumrequired bandwidth Bmin(objects/day)

Impact of subscription cost, c-10000 0 10000 20000 30000 40000 0 5000 10000 15000 20000 25000 30000c = 0.25c = 1.0c = 1.4c = 1.6Number of users, nUISP

Utility of ISP

Critical mass of users, nmin 0 20000 40000 60000 80000 100000 120000 140000 0 200 400 600 800 1000nminAverage object replication, r2 = 62 = 52 = 42 = 3Cost per unit of bandwidth

Model refinementsImpact of finite bandwidth of the users

Model refinementsImpact of finite bandwidth of the users

In case of constant traffic load, the cost for the ISP increases if it provides more download bandwidth to the users (!)The system performance is strongly affected by the upload bandwidth of the users, that should be larger than or equal to the download bandwidth (contrary to common practices, e.g., ADSL lines !)

Impact of asymmetric access bandwidths 0 2000 4000 6000 8000 10000 12000 0 500 1000 1500 2000 2500 3000 3500 4000Upload bandwidth, bubd = 0.1 buMinimumrequired bandwidth Bmin(objects/day)(for fixed number of users = 20000) bd = bubd = 2 bubd = 3 bubd = 4 bu

The impact of bd 0 1000 2000 3000 4000 5000 6000 7000 0 5000 10000 15000 20000 25000 30000bd = 2000Minimumrequired bandwidth Bmin(objects/day)Number of users, nbd = 1000bd = 500bd = 100bd = 10

ConclusionsWe have developed a simple analytical model to: analyze the interaction between P2P traffic and ISP understand the impact of several parametersguide the selection of the optimal strategy to manage P2P traffic

Future workMore model refinementsMultiple ISPs competing with each other

Questions ? Comments ?