1. 1 P2P Computing

2. 2 What is P2P? Server-Client model

3. 3 What is P2P? Traditional network architecture

4. 4 What is P2P? Problem with Server-Client Model  Scalability As the number of users increases, there is a higher demand for computing power, storage space, and bandwidth associated with the server-side  Reliability The whole network will depend on the highly loaded server to function properly

5. Why P2P? The Internet has three valuable fundamental assets  Information  Computing resources  Bandwidth

6. Why P2P? Hot Spots become hotter, Yet Cold spots keep cold Internet Personal Computers 80% idle CPU time Computers in our Lab 99% idle CPU time ! Yildiray’s Laptop 90% idle CPU time

7. 7 What is P2P? A peer in P2P network acts as both a client and a server in traditional client/server architecture Typical P2P file sharing application  Users publish the file they want to share  Users search the needed file by providing a query Query is sent to other online peers in the network  A peer that has local media files matching the query will return information on how to retrieve the files  Users may receive multiple successful responses and can select the files they want to retrieve The files are down loaded from the remote peer.

8. 8 query music category banner ad 3 million users online sharing 4 PetaBytes of data Kazaa Native Windows Application

9. 9 What is P2P? A pure P2P network architecture

10. 10 What is P2P? A hybrid P2P architecture

11. 11 Peer-to-Peer (P2P) Peer-to-Peer computing is inspired by the controversial music- sharing service Napster Instead of Internet information being held in a few central locations, Peer-to-Peer computing makes it theoretically possible to access the files and data residing on every personal computer connected to the Internet P2PNot P2P

12. 12 Napster, first widely used p2p-application The application: A p2p application for the distribution of mp3 files  Each user can contribute its own content How it works: Central index server  Maintains list of all active peers and their available content Distributed storage and download  Client nodes also act as file servers  All downloaded content is shared

13. 13 File Sharing: Napster Here is how Napster used to work:  It is first necessary to download and install Napster software and to sign up for a free account at Napster  By clicking the Napster icon, a connection is established with Napster Web site, where the name of the music searched can be entered in a search window  The Napster software tracks all users who are online at that particular time and provides access to tracks stored on users' hard drives  When the song requested is found, Napster establishes a connection between the two computers so that it could be downloaded  However, after the Recording Industry Association of America sued Napster for copyright infringements, the court ruled that Napster was indeed violating copyrights and it is shut down

14. 14 History, motivation and evolution - Napster (cont ’ d) Central index server … peers Initial join  Peers connect to Napster server  Transmit current listing of shared files to server join

15. 15 History, motivation and evolution - Napster (cont ’ d) 1) query 2) answer Content search  Peers sends song request to Napster server  Napster server checks song database and returns list of matched peers … peers Central index server

16. 16 History, motivation and evolution - Napster (cont ’ d) 1)request 2)download … peers File retrieval  The requesting peer contacts the peer having the file directly and downloads it Central index server 1)2)

17. 17 Napster was the first simple but successful P2P-appliciation. Many others followed… P2P File Download Protocols: 1999: Napster 2000: Gnutella, eDonkey 2001: Kazaa 2002: eMule, BitTorrent History, motivation and evolution - File Download

18. 18 P2P is not restricted to file download! P2P Protocols: 1999: Napster, End System Multicast (ESM) 2000: Gnutella, eDonkey 2001: Kazaa 2002: eMule, BitTorrent 2003: Skype 2004: PPLive Today:TVAnts, PPStream, SopCast… Next: Video-on-Demand, Gaming File Download Streaming Telephony Video-on- Demand Gaming Application type: History, motivation and evolution - Applications

19. 19 Why is P2P so successful? Scalable – It’s all about sharing resources  No need to provision servers or bandwidth  Each user brings its own resource Eliminating the single-source bottleneck

20. 20 Why is P2P so successful? (cont ’ d) Cheap - No infrastructure needed Everybody can bring its own content  Homemade content  Illegal content  But also legal content  … High availability – Content accessible most of time

21. 21 Problems Some of the providers of leading P2P applications earn revenue from third parties by embedding spyware into the applications A large amount of polluted or corrupted content has been published in file sharing systems Copyright problem Free-rider problem

22. 22 P2P-Overlay Build graph at application layer, and forward packet at the application layer It is a virtual graph  Underlying physical graph is transparent to the user  Edges are TCP connection or simply a entry of an neighboring node’s IP address The graph has to be continuously maintained (e.g. check if nodes are still alive)

23. 23 P2P-Overlay (cont ’ d) Underlay Overlay Source

24. Basic P2P elements Peers  Each peer has a unique Peer ID  Client  Server  Router Operations  Join/leave  File publish  File search 24

25. Principles of the P2P Paradigm Symmetric roles  Peers are function equally Scalable  The network and computing resources used at each peer exhibit a growth rate as a function of overlay size that is less than linear 25

26. Principles of the P2P Paradigm Autonomous  each peer determines when it joins/makes requests/leaves the overlay Resource sharing  Resource contribution should be mutually beneficial Resilient  P2P overlays are resilient in the fact of dynamic peer membership 26

27. 27 The P2P enabling technologies Unstructured p2p-overlays  Generally random overlay Structured p2p-overlays  Distributed Hash Tables (DHTs) Super nodes architecture

28. 28 Unstructured p2p-overlays Unstructured p2p-overlays do not really care how the overlay is constructed  Peers are organized in a random graph topology E.g., new node randomly chooses three existing nodes as neighbors  Build your p2p-service based on this graph Several proposals  Gnutella  KaZaA/FastTrack  BitTorrent

29. 29 One Example of usage of unstructured overlays Typical problem in unstructured overlays: How to do content search and query?  Flooding  Limited Scope, send only to a subset of your neighbors  Time-To-Live, limit the number of hops per messages Search “Britney Spears” Example of flooding: (similar to Gnutella) Found entry! NotifyUpload

30. 30 Unstructured p2p-overlays (cont ’ d) Unstructured p2p-overlays are just a framework, you can build many applications on top of it Unstructured p2p-overlays pros & cons  Pros Very flexible: copes with node churn Supports complex queries (conversely to structured overlays)  Cons There is a tradeoff between generated traffic (overhead) and the horizon of the partial view

31. 31 Structured p2p-overlays Motivation  Locate content efficiently Solution – DHT (Distributed Hash Table)  Particular nodes hold particular keys  Locate a key: route search request to a particular node that holds the key  Representative solutions Chord, CAN, Pastry/Tapestry, etc.

32. 32 Challenges to Structured p2p-overlays Load balance  spreading keys evenly over the nodes. Decentralization  no node is more important than any other. Scalability  Lookup must be efficient even with large systems Peer dynamics  Nodes may come and go, may fail  Ensure the node responsible for a key can always be found.