An Introduction to

Peer-to-Peer Networks
Presentation for
MIE456 - Information Systems
Infrastructure II

Vinod Muthusamy
October 30, 2003
 Agenda
n Overview of P2P
n Characteristics
n Benefits

n Unstructured P2P systems

n Napster (Centralized)
n Gnutella (Distributed)
n Kazaa/Fasttrack (Super-peers)

n Structured P2P systems (DHTs)

n Chord
n Pastry
n CAN

n Conclusions
 Client/Server Architecture
n Well known,
powerful, reliable Server
server is a data
source
Client Client
n Clients request data
from server Internet

n Very successful Client Client

model
n WWW (HTTP), FTP,
Web services, etc.

* Figure from http://project-iris.net/talks/dht-toronto-03.ppt

 Client/Server Limitations
n Scalability is hard to achieve
n Presents a single point of failure
n Requires administration
n Unused resources at the network edge

n P2P systems try to address these limitations

 P2P Computing*
n P2P computing is the sharing of computer resources and
services by direct exchange between systems.

n These resources and services include the exchange of

information, processing cycles, cache storage, and disk
storage for files.

n P2P computing takes advantage of existing computing

power, computer storage and networking connectivity,
allowing users to leverage their collective power to the
benefit of all.

* From http://www-sop.inria.fr/mistral/personnel/Robin.Groenevelt/
Publications/Peer-to-Peer_Introduction_Feb.ppt
 P2P Architecture
n All nodes are both
clients and servers
n Provide and consume Node
data
n Any node can initiate a Node Node
connection
Internet
n No centralized data
source
n The ultimate form of Node Node
democracy on the
Internet
n The ultimate threat to
copy-right protection on
the Internet

* Content from http://project-iris.net/talks/dht-toronto-03.ppt

 P2P Network Characteristics
n Clients are also servers and routers
n Nodes contribute content, storage, memory, CPU
n Nodes are autonomous (no administrative
authority)
n Network is dynamic: nodes enter and leave the
network frequently
n Nodes collaborate directly with each other (not
through well-known servers)
n Nodes have widely varying capabilities
 P2P Benefits
n Efficient use of resources
n Unused bandwidth, storage, processing power at the edge of the network

n Scalability
n Consumers of resources also donate resources
n Aggregate resources grow naturally with utilization

n Reliability
n Replicas
n Geographic distribution
n No single point of failure

n Ease of administration
n Nodes self organize
n No need to deploy servers to satisfy demand (c.f. scalability)
n Built-in fault tolerance, replication, and load balancing
 P2P Applications
n Are these P2P systems?

n File sharing (Napster, Gnutella, Kazaa)

n Multiplayer games (Unreal Tournament, DOOM)

n Collaborative applications (ICQ, shared whiteboard)

n Distributed computation (Seti@home)

n Ad-hoc networks
 Popular P2P Systems
n Napster, Gnutella, Kazaa, Freenet

n Large scale sharing of files.

n User A makes files (music, video, etc.) on their
computer available to others
n User B connects to the network, searches for
files and downloads files directly from user A

n Issues of copyright infringement

 Napster
n A way to share music files with
others

n Users upload their list of files to

Napster server
n You send queries to Napster
server for files of interest
n Keyword search (artist, song,
album, bitrate, etc.)
n Napster server replies with IP
address of users with matching
files
n You connect directly to user A
to download file

* Figure from http://computer.howstuffworks.com/file-sharing.htm

 Napster
n Central Napster server
n Can ensure correct results
n Bottleneck for scalability
n Single point of failure
n Susceptible to denial of service
n Malicious users
n Lawsuits, legislation

n Search is centralized
n File transfer is direct (peer-to-peer)
 Gnutella
n Share any type of files
(not just music)
n Decentralized search
unlike Napster

n You ask your neighbours

for files of interest
n Neighbours ask their
neighbours, and so on
n TTL field quenches
messages after a
number of hops
n Users with matching files
reply to you

* Figure from http://computer.howstuffworks.com/file-sharing.htm

 Gnutella
n Decentralized
n No single point of failure
n Not as susceptible to denial of service
n Cannot ensure correct results

n Flooding queries
n Search is now distributed but still not scalable
 Kazaa (Fasttrack network)
n Hybrid of centralized Napster and decentralized Gnutella

n Super-peers act as local search hubs

n Each super-peer is similar to a Napster server for a small portion of
the network
n Super-peers are automatically chosen by the system based on their
capacities (storage, bandwidth, etc.) and availability (connection
time)

n Users upload their list of files to a super-peer

n Super-peers periodically exchange file lists
n You send queries to a super-peer for files of interest
 Free riding*
n File sharing networks rely on users sharing data

n Two types of free riding

n Downloading but not sharing any data
n Not sharing any interesting data

n On Gnutella
n 15% of users contribute 94% of content
n 63% of users never responded to a query
n Didnt have interesting data

* Data from E. Adar and B.A. Huberman (2000), Free Riding on Gnutella
 Anonymity
n Napster, Gnutella, Kazaa dont provide
anonymity
n Users know who they are downloading from
n Others know who sent a query

n Freenet
n Designed to provide anonymity among other
features
 Freenet
n Data flows in reverse path of query
n Impossible to know if a user is initiating or forwarding a query
n Impossible to know if a user is consuming or forwarding data

n Smart queries
n Requests get
routed to
correct peer
by
incremental
discovery
 Structured P2P
n Second generation P2P overlay networks

n Self-organizing
n Load balanced
n Fault-tolerant

n Scalable guarantees on numbers of hops to answer a

query
n Major difference with unstructured P2P systems

n Based on a distributed hash table interface

 Distributed Hash Tables (DHT)
n Distributed version of a hash table data structure
n Stores (key, value) pairs
n The key is like a filename
n The value can be file contents

n Goal: Efficiently insert/lookup/delete (key, value) pairs

n Each peer stores a subset of (key, value) pairs in the
system
n Core operation: Find node responsible for a key
n Map key to node
n Efficiently route insert/lookup/delete request to this node
 DHT Generic Interface
n Node id: m-bit identifier (similar to an IP address)
n Key: sequence of bytes
n Value: sequence of bytes

n put(key, value)
n Store (key,value) at the node responsible for the key
n value = get(key)
n Retrieve value associated with key (from the
appropriate node)
 DHT Applications
n Many services can be built on top of a DHT
interface
n File sharing
n Archival storage
n Databases
n Naming, service discovery
n Chat service
n Rendezvous-based communication
n Publish/Subscribe
 DHT Desirable Properties
n Keys mapped evenly to all nodes in the
network
n Each node maintains information about only
a few other nodes
n Messages can be routed to a node
efficiently
n Node arrival/departures only affect a few
nodes
 DHT Routing Protocols
n DHT is a generic interface

n There are several implementations of this interface

n Chord [MIT]
n Pastry [Microsoft Research UK, Rice University]
n Tapestry [UC Berkeley]
n Content Addressable Network (CAN) [UC Berkeley]

n SkipNet [Microsoft Research US, Univ. of Washington]

n Kademlia [New York University]
n Viceroy [Israel, UC Berkeley]
n P-Grid [EPFL Switzerland]
n Freenet [Ian Clarke]

n These systems are often referred to as P2P routing substrates or P2P

overlay networks
 Chord API
n Node id: unique m-bit identifier
(hash of IP address or other unique ID)
n Key: m-bit identifier (hash of a sequence of bytes)
n Value: sequence of bytes

n API
n insert(key, value) store key/value at r nodes
n lookup(key)
n update(key, newval)
n join(n)
n leave()
 Chord Identifier Circle
n Nodes organized in
an identifier circle
based on node
identifiers
n Keys assigned to
their successor node
in the identifier circle
n Hash function
ensures even
distribution of nodes
and keys on the
circle
 Chord Finger Table
n O(logN)
table size

n ith finger
points to
first node
that
succeeds n
by at least
2i-1
 Chord Key Location
n Lookup in
finger table
the furthest
node that
precedes
key

n Query
homes in on
target in
O(logN)
hops
 Chord Properties
n In a system with N nodes and K keys, with
high probability
n each node receives at most K/N keys
n each node maintains info. about O(logN) other
nodes
n lookups resolved with O(logN) hops

n No delivery guarantees
n No consistency among replicas
n Hops have poor network locality
 Network locality
n Nodes close on ring can be far in the
network.
To vu.nl
Lulea.se

OR-DSL N20
CMU
MIT
MA-Cable
Cisco

CA-T1
N40 Cornell

N41
CCI

N80
NYU
Aros
Utah

* Figure from http://project-iris.net/talks/dht-toronto-03.ppt

 Pastry
n Similar interface to Chord

n Considers network locality to

minimize hops messages
travel

n New node needs to know a

nearby node to achieve
locality

n Each routing hop matches

the destination identifier by
one more digit
n Many choices in each hop
(locality possible)
 CAN
n Based on a
d-dimensional
Cartesian
coordinate
space on a
d-torus

n Each node
owns a distinct
zone in the
space
n Each key
hashes to a
point in the
space
CAN Routing and Node Arrival
 P2P Review
n Two key functions of P2P systems
n Sharing content
n Finding content

n Sharing content
n Direct transfer between peers
n All systems do this
n Structured vs. unstructured placement of data
n Automatic replication of data

n Finding content
n Centralized (Napster)
n Decentralized (Gnutella)
n Probabilistic guarantees (DHTs)
 Conclusions
n P2P connects devices at the edge of the Internet

n Popular in industry
n Napster, Kazaa, etc. allow users to share data
n Legal issues still to be resolved

n Exciting research in academia

n DHTs (Chord, Pastry, etc.)
n Improve properties/performance of overlays

n Applications other than file sharing are being developed