A peer-to-peer

In client-server architectures:

• servers are managed by the service providers and become bottlenecked when the number of clients grows
• administration and fault recovery costs dominate
• the network bandwidth available to a server site is also a major constraint

In peer-to-peer architectures, support distributed services and applications using data and computing resources available in the personal computers and workstations

p2p exploits resources available at the "edges" of the internet [Shirky 2000]

The key aspects of peer-to-peer architectures are the placement and retrieval of information objects.

System Characteristics

• design ensures each user contributes resources to the system
• although participants may differ in the resources that they contribute, all the nodes have the same functional capabilities and responsibilities
• correct operation doesn't depend on the existence of any centrally administered systems
• can be designed to offer a limited degree of anonymity to the providers and users of resources
• a key issue for the efficient operation is the choice of an algorithm for the placement of data across many hosts and subsequent access to it in a manner that balances the work load and ensures availability without adding undue overhead

Examples

Early Systems

• DNS and Usenet adopted multi-server system
• Xerox Grapevine name registration and main delivery system
• SETI@home
  • Search for Extra-Terrestrial Intelligence
  • a few servers and lots of client machines/volunteers
  • distribution of tasks managed by servers, but work completed by users

Modern Systems

• First Generation: Napster
• Second Generation: file sharing applications
  • Freenet, Gnutella, Kazaa, BitTorrent
• Third Generation: distributed object store
  • also known as DHT
  • sometimes application agnostic middleware
  • examples: Chord, Pastry, Tapestry, Kademlia, Content-Addressable Network (CAN)
• Fourth Generation: blockchain

P2P Systems

• distribution
  • place resources across different peers
• discovery / routing
  • when requested, messages are routed to discover where the resources are and retrieve them (routing)
• replication
  • resources are typically replicated onto multiple peers for better availability
• naming
  • resources have names or IDs (usually a GUID)

P2P systems are usually good for immutable objects.

P2P Middleware

Key issue: to distribute resources and locate them. in a decentralized fashion.