GridComputing9-1经济学原理及其在网格中的应用.pptVIP

Diversity of P2P Systems Distributed storage, search, and retrieval File-sharing: Napster, gnutella, kaZaA, Overnet, BT … Anonymity/Persistence: Eternity, Freehaven, FreeNet, Publius, … DHTs: Chord, CAN, Pastry, Tapestry, OpenHash, … Distributed computation Globus (grid), Entropia, SETI@Home, etc. 2007-4-26 * * Diversity of P2P Systems 2007-4-26 * * P2P System Characteristics What do P2P systems have in common? No infrastructure or service provider: rely on contributions by individual peers Hidden action: difficult to monitor or enforce cooperation Ad-hoc communities: highly dynamic memberships; interactions with strangers 2007-4-26 * * Free-riding Fundamental tension between individual rationality and collective welfare System utility derived solely from peer contributions Contributions not costless→ disincentives to share Rational peers choose to free-ride, i.e., consume but not contribute 2007-4-26 * * Free-riding Free-riding prevalent in file-sharing networks [Adar00; Sariou02] 66% of gnutella peers share no files 10% of peers share 87% of files 20% of peers share 98% of files [Adar00]: “Tragedy of digital commons”? 2007-4-26 * * Questions What are the costs of participating in a P2P network? How significant are the disincentives for sharing (potential legal liability notwithstanding)? What are the effects of free-riding on P2P system performance? Are P2P systems doomed to failure due to non-cooperation? 2007-4-26 * * Questions How do we design incentive mechanisms to encourage cooperation in P2P systems? 2007-4-26 * * Disincentive for Sharing Case 1: P2P file-sharing Incoming link utilization degrades by 20-80% when simultaneously uploading (ns-2 simulation) Case 2: P2P media streaming Streaming quality becomes highly variable as uploading bandwidth increases (planetlab experiment using PROMISE prototype) 2007-4-26 * * Participation Cost Cost can be highly variable, dependent on many factors, e.g., item popularity, network topology, rout