A Modeling Framework to Understand the Tussle between ISPs and Peer-to-Peer File Sharing Users Michele Garetto – Univ. of T orino Daniel Figueiredo – Fed. Univ. of Rio de Janeiro Rossano Gaeta – Univ. of T orino Matteo Sereno – Univ. of T orino IFIP Performance 2007
An Internet Tale Once upon a time... Internet ISP user user unhappy (“world wide wait”) ISP unhappy (little revenue) Then came broadband access... fast! Internet ISP user And they lived happily ever after... IFIP Performance 2007
The Villain Arrives P2P file sharing application (Kazza, Bittorrent, Emule, etc) ISPs hate it! users love it! users using good and free ISP their link content, overnight user Internet link downloads utilization gone wild degrades all subscribers ISP Internet more bandwidth costs money! IFIP Performance 2007
Taking Care of The Villain seriously threaten Mafia application developers! style! doesn't seem to work (Napster story) Is it Really a Villain? Users love it! Driving force for broadband adoption Increased revenue for ISPs user IFIP Performance 2007
Some Other Options User unfriendly ideas User friendly ideas increase acquire more user subscription cost bandwidth volume based network caching pricing application-layer block / shape P2P redirection traffic What should the ISP do? ??? ISP IFIP Performance 2007
The Real Thing (Data) P2P represented 60% of Internet Traffic at the end of 2004! IFIP Performance 2007
Our contribution Modeling framework to analyze interactions between P2P file sharing users (their traffic) and ISP economic + performance models Basic insights about system dynamics Used to evaluate different strategies to manage P2P traffic IFIP Performance 2007
Meet the Players generates goal : to make ISP queries money! quality of service sets subscription user expectations price what's hot, controls what's not bandwidth influences P2P P2P application app. behavior locates object network architecture Network protocols IFIP Performance 2007
System Setting number of P2P users outside ISP User issues number of P2P query users within ISP B d constrained resource for ISP Outside download consumes B d IFIP Performance 2007
Simple System Model unconstrained downloads from prob.object is within the ISP prob. P2P app. located inside ISP locates object aggregate query rate Model for “Internet to ISP” link object retrieval probability: IFIP Performance 2007
User Utility Function Satisfaction model for user i user probability of shape subscription successful parameter cost object retrieval User utility 1 0.5 Minimum 1 σ 0 service level for 0 0.2 0.4 0.6 0.8 user i -0.5 -1 IFIP Performance 2007
ISP Utility Function Profit for ISP (revenue - costs) ISP cost per fixed revenue from unit of charge subscribers’ fee external bandwidth The ISP starts service only if IFIP Performance 2007
Modeling Traffic Locality Probability there exis t at least one internal replica of object replicated r times in the system Number of internal copies Number of external copies Probability to download from internal replica locality parameter IFIP Performance 2007
Analytical Results How much bandwidth should the ISP buy to minimally satisfy the users? B min = max [ 0, n q min − q r n / N ] identical users and n >> N Non-linear behavior (on n ) more users, more locality, less BW needed can be zero if n large enough May not yield profit too few users, too costly to satisfy them Dependent on multiple parameters IFIP Performance 2007
Impact of Object Replication (r) 5000 r = 500 B min (objects/day) more bw needed r = 1000 to support larger 4000 user population r = 1500 3000 less bw needed (users satisfied locally) 2000 1000 0 0 10000 20000 30000 40000 Number of users, n more replicas, better locality, lower B min IFIP Performance 2007
Impact of Subscription Cost (c) 40000 c = 0.25 c = 1.0 Utility of ISP 30000 n min c = 1.4 20000 c = 1.6 10000 ISP does 0 not provide service! -10000 0 5000 10000 15000 20000 25000 30000 Number of users, n critical mass of users, n min lower cost, more profit earlier, less profit later IFIP Performance 2007
Critical Mass of Users, n min 140000 β 2 = 6 β 2 = 5 120000 Cost per β 2 = 4 unit of 100000 bandwidth β 2 = 3 for ISP 80000 n min 60000 40000 20000 0 0 200 400 600 800 1000 Average object replication, r higher bw cost for ISP, higher critical mass large influence of number of replicas IFIP Performance 2007
Model Refinements Simple model Refined model relax these users' access assumptions bandwidth are unconstrained propose object object replication popularity and is a parameter replication model all objects are identical (no popularity) users availability identical IFIP Performance 2007
Object Popularity and Replication Model Temporal evolution of object popularity Objects' popularities evolve differently Objects continuously introduced and removed by users Number of replicas of an object at time t? Analytical technique based on Poisson shot noise process IFIP Performance 2007
Example A video from the popularity object news A popular song t 1 t 2 time object request at request time, both have same popularity, but news has more replicas IFIP Performance 2007
Limited Bandwidth Refinements each user within ISP modeled separately users upload rate of download bandwidth requests to user i limited to b u users BW consumption is limited to b d IFIP Performance 2007
Results from Refined Model 7000 b d = 2000 b d = 1000 6000 b d = 500 b d = 100 Degenerate to 5000 b d = 10 B min 4000 simple model 3000 2000 when parameters 1000 set appropriately 0 0 5000 15000 25000 Number of users, n Other interesting insights influence of limited upload bandwidth upload/download bandwidth asymmetry object popularity and replication influence of user impatience IFIP Performance 2007
Impact of asymmetric access bandwidths (for fixed number of users = 20000) 12000 b d = b u B min (objects/day) 10000 b d = 2 b u b d = 3 b u 8000 b d = 4 b u 6000 4000 2000 0 0 500 1000 1500 2000 2500 3000 3500 4000 Upload bandwidth, b u cost for ISP increases as ratio increases better if upload BW is greater than download IFIP Performance 2007
Conclusions Development of simple analytical model economics + performance interaction between P2P users (their traffic) and ISP insights into strategy for ISP to manage its traffic Model for object popularity and replication of independent interest Future work Multiple ISPs competing with each other IFIP Performance 2007
THE END Thank you! Questions? Comments? IFIP Performance 2007
CacheLog ic Impact on Service Providers Advanced Solutions for P2P Networks P2P is driving consumer broadband uptake P2P is driving consumer broadband uptake …and broadband is driving P2P uptake and broadband is driving P2P uptake … P2P is the the dominant protocol In excess of 92% of P2P traffic crosses transit/peering links P2P protocols will aggressively consume aggressively consume any available bandwidth capacity Due to P2P’s symmetrical nature on average 80% of upstream capacity is consumed by P2P P2P affects QoS levels for ALL subscribers Service Providers can not afford to block or restrict P2P ISPs must intelligently manage P2P - blocking and shaping doesn’t work BitT orrent HT T P eDonkey O ther Non P2P FastT rack O ther P2P Gnutella Recognising Presentation | WCW 2005
The ISP perspective vs P2P: threat or opportunity ? P2P traffic: friend or foe ? friend: driving force for adoption of broadband access by the users foe: 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 ? IFIP Performance 2007
Strategies to manage P2P traffic Acquire more bandwidth Block P2P traffic Traffic shaping (e.g., priority to non-P2P) Pricing schemes based on user traffic volumes / bandwidth caps Network caching / customized P2P application within ISP Application-layer redirection of P2P traffic IFIP Performance 2007
Results Assumption: n identical users N = 50 millions request rate by user (object/day) introduction of new contents by user (object/day) Minimum required external bandwidth: IFIP Performance 2007
The impact of efficacy in exploiting traffic locality ( ) 100000 Minimum = 0.25 required = 0.5 80000 bandwidt = 0.75 h = 1.0 60000 B min 40000 (objects/day) 20000 0 0 100000 300000 500000 700000 Number of users, n IFIP Performance 2007
Recommend
More recommend
