150 likes | 326 Views
P4P: Towards Cooperation between P2P and ISPs. Haiyong Xie (Yale) Arvind Krishnamurthy (U. Washington) Avi Silberschatz (Yale) Y. Richard Yang (Yale). 2007-7-25. http://www.cachelogic.com. P2P Content Distribution.
E N D
P4P: Towards Cooperation between P2P and ISPs Haiyong Xie (Yale) Arvind Krishnamurthy (U. Washington) Avi Silberschatz (Yale) Y. Richard Yang (Yale) 2007-7-25
http://www.cachelogic.com P2P Content Distribution • Traffic volume: up to 60-70% of Internet traffic is contributed by P2P applications [cachelogic] • Traffic pattern: random peering (e.g., BitTorrents) causes traffic spread across PoPs and domains • Problems • Increased network resource usage (e.g., using bandwidth of more links) • Increased network operational costs • Degraded performance to other applications P4PWG July Meeting
ISPs try to “manage” P2P traffic P2P tries to evade from being captured • Upgrade network infrastructure • Deploy P2P caching devices • Terminate connectivity • Rate limit P2P traffic • Uses random ports • Encrypts traffic Bandwidth Battle between ISPs and P2P • The battle results in a lose-lose situation P4PWG July Meeting
Where is the Fundamental Problem? • Traditional ISP application feedback/control: • Routing/traffic engineering (TE) • Rate control through congestion feedback (packet drops) • Ineffective for P2P • due to highly dynamic, scattered traffic pattern caused by dynamic, unguided (network-oblivious) peer selection Objective: design a framework to enable better ISP and P2P cooperation P4PWG July Meeting
Design Rationale • Performance improvement • for both ISPs and P2P • Scalability • support a large number of P2P users and networks in dynamic settings • Privacy preservation • Extensibility • Application-specific requirements • Tracker-based vs. trackerless P2P systems • Gossip among peers • Incremental deploymentability • ISP contribution for P2P acceleration P4PWG July Meeting
The P4P Framework • P4P: proactive provider participation for P2P; P2P for providers; provider portal for P2P, … • Two components • Control plane • iTrackers: an ISP portal for P2P and content providers • Three levels: • Network status:e.g., network topology • ISP policy and guideline: e.g., traffic balance ratio for inter-AS peering links, time of day preference • ISP capabilities: e.g., QoS, CoS, ISP servers participation in content distributions • Data plane [optional] • Routers on data paths provide fine-grained, ISP policy-based feedbacks, e.g., utilize TCP ECN bit or feedback fields in an overlay packet header P4PWG July Meeting
appTracker iTracker B 1 1 2 2 iTracker A 3 3 [4] b ISP A ISP B a P4P Control Path: Obtain Network Status/Policy An iTracker for each ISP. 1: Peer queries iTracker of local ISP to obtain network status/policy 2,3: Tracker-based: peer reports status/policy to appTracker; appTracker selects peering set considering both ISP status/policy and application requirements [4]: Trackerless: peers exchange information and make peering decisions P4PWG July Meeting
P4P Control Path : Request Capability 6 appTracker 5 iTracker B iTracker A b a appTracker/content provider requests ISP capabilities to accelerate content distribution. 7 ISP A ISP B 5: appTracker [content provider] requests ISP B’s participation in content distribution 6: ISP B allocates servers to accelerate content distribution 7: appTracker includes ISP B’s servers in returned peering sets to peers Note: this can be extended to handle trackerless systems, as we did in the previous slide 2007-7-25 8 P4PWG July Meeting
P4P Framework: Data Path [optional] ISP A ISP B b a Routers mark packets to provide faster, fine-grained feedbacks, e.g., virtual capacity to optimize multihoming cost and performance Peers adjust traffic rates according to feedbacks 2007-7-25 9 P4PWG July Meeting
Test Plan for P4P • Measurement study with Pando (in progress) • Evaluate P2P self-adaptation schemes (in progress) • Generate best practices for P2P design • Serve as comparison basis • iTracker (in progress) • Network information (roughly completed) • ISP policy and guideline (in progress) • ISP capability (in progress) • Data path (in progress) • Evaluate P4P design with Pando and Verizon (in progress) P4PWG July Meeting
Preliminary Results • Simulations • Discrete-event simulation • a module for modeling BitTorrent protocol • a module for modeling underlying network topology and data transfer dynamics using TCP rate equation • Network topology: PoP-level AT&T and Abilene topologies • Network routing: OSPF routing • Internet experiments • 53 Internet2 nodes on PlanetLab • iTracker for Abilene network • Use OSPF routing to re-construct traffic load on Abilene links P4PWG July Meeting
Evaluation – BitTorrent on Abilene • Compared to P4P, native P2P can result in • 2x download completion time • 2x higher link utilization • Native P2P can result in some peers experiencing very long download completion time • Native P2P can result in much larger variance in link utilization P4PWG July Meeting
Evaluation – BitTorrent on AT&T • Compared to P4P, native P2P can result in • 1.6x download completion time • 3x higher link utilization • Some peers can experience very long download completion time with native P2P • Link utilization variance can be larger for native P2P P4PWG July Meeting
Evaluation – Liveswarms on PlanetLab • Liveswarms* is a P2P-based video streaming application, which adapts BitTorrent protocol to video streaming context • Run liveswarms on 53 PlanetLab nodes for 900 seconds • P4P and native liveswarms achieve roughly the same amount of throughput • P4P reduces link load • Average link load saving is 34MB • Maximum average link load saving is 60% • Native liveswarms:1Mbps • P4P liveswarms: 432Kbps *Michael Piatek, Colin Dixon, Arvind Krishnamurthy, Tom Anderson. LiveSwarms: Adapting BitTorrent for end host multicast. Technical report: UW-CSE-06-11-01 P4PWG July Meeting
Contact and Acknowledgement For further details, please refer to our technical report Yale/DCS TR-1377 It is still a work-in-progress and changes rapidly Questions/comments are highly welcome: Haiyong Xie (haiyong.xie@yale.edu) Y. Richard Yang (yry@cs.yale.edu) Acknowledgements We would like to thank Charles Kalmanek (AT&T Labs), Marty Lafferty (DCIA), Doug Pasko (Verizon), Laird Popkin (Pando), Keith Ross (Polytechnic), Ke Xu (Tsinghua Univ.) for suggestions, discussions and feedback. 2007-7-25 15 P4PWG July Meeting