1. ADVANCED COMPUTER NETWORKS Peer-Peer (P2P) Networks 1

2. What is P2P?  “the sharing of computer resources and services by direct exchange of information” 2

3. What is P2P?  Various definitions seem to agree on  sharing of resources  direct communication between equals (peers)  no centralized control 3

4. What is a peer?  “…an entity with capabilities similar to other entities in the system.” 4

5. Client/Server Architecture  Well known, powerful, reliable server is a data source  Clients request data from server  Very successful model  WWW (HTTP), FTP, Web services, etc. Server Client Internet 5

6. Client/Server Limitations  Scalability is hard to achieve  Presents a single point of failure  Requires administration  P2P systems try to address these limitations 6

7. P2P Architecture  All nodes are both clients and servers  Provide and consume data  Any node can initiate a connection  No centralized data source  “The ultimate form of democracy on the Internet”  “The ultimate threat to copy- right protection on the Internet” Node Internet 7

8. P2P Network Characteristics  Clients are also servers and routers  Nodes contribute content, storage, memory, CPU  Nodes are autonomous (no administrative authority)  Network is dynamic: nodes enter and leave the network “frequently”  Nodes collaborate directly with each other (not through well- known servers) 8

9. P2P Applications  File sharing (BitTorrent, Napster, Kazaa)  Multiplayer games (Counter Strike, Unreal Tournament, DOOM)  Collaborative applications (ICQ, shared whiteboard) 9

10. Popular file sharing P2P Systems  BitTorrent  Large scale sharing of files.  User A makes files (music, video, etc.) on their computer available to others  User B connects to the network, searches for files and downloads files directly from user A  Issues of copyright violation 10

11. Approaches  Structured  Unstructured 11

12. Structured original “Napster” design - when peer connects, it informs central server:  IP address  content 2) User B searches for file 3) User B requests file from User A centralized directory server peers User B User A 1 1 1 1 2 3 12

13. Centralized model AB DC file transfer is decentralized, but locating content is highly centralized 13

14. Napster: how does it work Four steps:  Connect to Napster server  Upload your list of files to server.  Give server keywords to search the full list with.  Select “best” of correct answers. 14

15. Napster napster.com users File list is uploaded 1. 15

16. Napster napster.com user Request and results User requests search at server. 2. 16

17. Napster napster.com user pings User pings hosts that apparently have data. 3. 17

18. Napster napster.com user Retrieves file User retrieves file 4. 18

19. Unstructured  Fully distributed  no central server  Used by KaZaa  Each peer indexes the files it makes available for sharing (and no other files) 19

20. KaZaa: Ping flooding Query QueryHit Query QueryHit Query QueryHit File transfer: HTTP r Ping message sent over existing TCP connections r peers forward ping message r Pong sent over reverse path 20

21. KaZaa  file divided into 256KB chunks.  while downloading, peer uploads chunks to other peers.  peers may come and go  once peer has entire file, it may leave or remain 21

22. P2P : Skype Skype clients (SC) Supernode (SN) 22

23. Issues with P2P  Free Riding Downloading but not sharing any data  On BitTorrent 15% of users contribute 94% of content 63% of users never responded to a query  No ranking: what is a trusted source? 23