Interoperability of PeerToPeer File Sharing Protocols

1
Interoperability of Peer-To-Peer File Sharing
Protocols

• Jeffrey Zemerick

2
Outline

• Introduction and Overview of P2P Communication
• Peer-to-Peer Protocols (Gnutella, Napster,
  Freenet)
• Interoperability of P2P Protocols
• Potential Problems
• Conclusion

3
Introduction

• The goal is to access multiple peer-to-peer (P2P)
  networks from the same client.
• Doing so would provide access to many, many new
  peers.
• Proposed solution has advantages and
  disadvantages.

4
Overview of P2P Communication

• Two main types Pure P2P and Server-Mediated P2P
• In Server-Mediated P2P, a central server
  coordinates the peers.
• If the central server should fail, the entire
  network will be unavailable (Napster).

Server-mediated P2P
5
Overview of P2P Communication

• In a Pure P2P network there is no central server.
• New peers connect to the network either by
  flooding broadcasts, or by connecting to a
  superpeer (or supernode).
• The failure of any single node will not affect
  the availability of the entire network.

Pure P2P
6
Outline

• Introduction and Overview of P2P Communication
• Peer-to-Peer Protocols (Gnutella, Napster,
  Freenet)
• Interoperability of P2P Protocols
• Potential Problems
• Conclusion

7
Peer-to-Peer Protocols Gnutella

• Pure P2P network.
• First Gnutella client was available in 2000.
• Developed by Nullsoft (now AOL).
• Currently available in several versions 0.4,
  0.6, and Gnutella2.
• 0.6 is an upgrade to 0.4, however, Gnutella2 is
  considered a completely new protocol opposed to
  an upgrade.

8
Peer-to-Peer Protocols Gnutella

• Peers in Gnutella 0.6 are divided into two groups
  peers and superpeers.
• In Gnutella 0.4, all peers essentially acted as
  superpeers.

Gnutella 0.6 network.
9
Peer-to-Peer Protocols Freenet

• Pure P2P network.
• Developed by Ian Clarke as a means to promote
  free speech by providing users true anonymity.
• Files are broken into pieces, encrypted, and then
  stored separately on different computers
  impossible to tell what files are being shared by
  what users.

10
Peer-to-Peer Protocols Napster

• Server-mediated P2P network.
• Launched in 1999, promptly shut down in 2001.
• Unlike the others, Napster requires a user to log
  in (username, password) to the network first.
• Search requests go through central server, file
  transfers done between users.

11
Other Popular P2P Protocols

• FastTrack (Kazaa) Pure P2P network. Uses the
  weak (but fast) UUHash hashing algorithm.
• BitTorrent Pure P2P network to let users share
  large files.
• eDonkey Pure P2P network for audio files no
  longer officially available due to the RIAA.
  eDonkey client and eDonkey server is two separate
  programs.

12
Outline

• Introduction and Overview of P2P Communication
• Peer-to-Peer Protocols (Gnutella, Napster,
  Freenet)
• Interoperability of P2P Protocols
• Potential Problems
• Conclusion

13
Interoperability of the P2P Protocols

• All the P2P protocols are different, however,
  their messages are all similar.
• By knowing the messages of each protocol we can
  map each message(s) to its corresponding
  message(s) in the other two protocols.

14
Messages in the P2P Protocols
Napster Messages
Freenet Messages
Gnutella Messages
15
Solution Gateway

• Authors of the paper suggest implementing a
  gateway that translates messages from one P2P
  network to the other P2P networks.
• Could be implemented in either hardware or
  software (similar to a proxy server).
• Both implementation methods have advantages and
  disadvantages.

16
Solution Gateway
Software Gateway
Hardware Gateway
17
Potential Problems Malware

• Each P2P network are littered with viruses,
  spyware, and other malicious software.
• Accessing more than one network at a time
  essentially triples the risk of acquiring
  malware.
• Adding a virus scanner to the gateway may help
  prevent the transmission of malware.

18
Potential Problems Network Logins

• Some P2P networks (Napster) require the user to
  log in to the network first.
• The gateway must be aware of these logins
  (username, password combinations).
• (Pure P2P networks do not require a login.)

19
Potential Problems MD5 Hash

• The authors propose using the MD5 hash algorithm
  to create a file that uniquely identifies each
  peer.
• MD5 is known to be weak and it is now well-known
  that two files can be constructed to have the
  same MD5 hash.
• A different hash algorithm should be used.

20
Positive Aspects

• A gateway does not require any modification of
  the client software.
• (A software-based gateway would require the user
  to adjust the clients proxy server settings.)

21
Conclusion

• A gateway is an effective solution to access
  multiple P2P networks.
• However, there are P2P clients that have the
  ability to connect to multiple P2P networks
  built-in
• iMesh can access the FastTrack, eDonkey,
  Gnutella, and Gnutella2 networks.