Introduction of PeerToPeer Networks

1
Introduction of Peer-To-Peer Networks

• Vesile Evrim

2
Outline

• What is a peer to peer system
• Peer Oriented Applications
• Peer Oriented Systems
• P2P Challenges
• Search
• Resource Management
• Security Privacy

3
What is P2P?
pastry
jxta
can
fiorana
napster
freenet
united devices
open cola
?
aim
ocean store
netmeeting
farsite
gnutella
icq
ebay
maorpheus
limewire
seti_at_home
bearshare
uddi
grove
jabber
popular power
kazaa
folding_at_home
tapestry
mojo nation
process tree
chord
4
What is a P2P System?

• Multiple sites (at edge)
• Distributed resources
• Sites are autonomous (different owners)
• Sites are both clients and servers
• Sites have equal functionality

P2P Purity
5
Introduction to Peer-to-Peer

• Evolution of the Internet

6
P2P Benefits

• Pooling availible(inexpensive) resources
• High availibilty and falut- tolerance
• Self Organiaztion
• Scalable

7
Peer-oriented Applications

• File Sharing document sharing among peers with
  no or limited central controls. E.g., Napster,
  Gnutella, Freenet, KaZaA
  
• Instant Messaging (IM) Immediate voice and file
  exchanges among peers. AOL Messenger, MSN, ICQ
  
• Distributed Processing One can widely utilize
  resources available in other remote peers. E.g.,
  SETI_at_Home, Folding_at_Home,

8
What is SETI_at_Home?

• is a project that allows millions around the
  world to participate in the search for
  extraterrestrial life.
  
• distribute a free screensaver
• uses the computing power of millions of personal
  computers worldwide to process data collected at
  the Arecibo radio telescope in Puerto Rico.
  
• let personal computer users participate in
  processing data that might come from an
  extraterrestrial source.  
  

9
Peer-oriented Systems

• Structured P2P P2P network topology is tightly
  controlled and files are placed not at random
  nodes but at specified locations that will make
  subsequent queries easier to satisfy. Examples
  CAN, Chord, Pastry
• Unstructured P2P no precise control over the
  network topology or file placement. The placement
  of files is not based any knowledge of the
  topology. Examples Napster, Gnutella, KaZaA

10
Unstructured P2P

• Centralized Model Napster
• Decentralized Model Gnutella
• Hierarchical Model KaZaA

11
NAPSTER (Centralized Model)

• Central indexing and searching
• Retrieval four steps
• Connect to Napster server
• Upload your list of files (push) to server.
• Give server keywords to search the full list
  with.
  
• Select best of correct answers. (pings)
• no security passwords in plain text, no
  authentication, no anonymity
  

12
Napster How it works?(1)
File list is uploaded
1.
Napster.comcom
users
13
Napster How it works?(2)
User requests search at server.
2.
Napster.com
Request and results
user
14
Napster How it works?(3)
User pings hosts that apparently have data.

Looks for best transfer rate.
3.
napster.com
pings
pings
user
15
Napster How it works?(4)
User retrieves file
4.
napster.com
Retrieves file
user
16
Decentralized ModelGNUTELLA(Query Flooding)

• Peer to Peer indexing and searching service
• Send query to neighbors
• Neighbors forward query
• If queried peer has object, it sends message back
  to querying peer
  
• TTL is mechanism to limit query distance.
• TTL(Time To Live) the number of times a query
  will be forwarded before it is removed from the
  network.
  

17
Gnutella (cont)
query
18
Hierarchical ModelKaZaA

• Each peer is either a supernode or assigned to a
  supernode.
  
• Supernode tracks the content in all its
  children.
  
• Peer queries supernode, supernode may query other
  supernodes (flooding).
  

19
Freenet

• Peer-to-peer indexing and searching service.
• Peer-to-peer file downloading.
• Data is stored in certain peers
• More Privacy
• Files served use the same route as searches (not
  point-to-point)
  

20
CAN (Structured P2P)

• Divide space logically into dimensions
• Hash functions are used to place peers and
  resources
  
• Lose autonomity
• More efficient search

Nodes
Data
21
P2P Challenges

• Search
• Resource Management
• Security Privacy

22
Search

• Query Expressiveness
• Comprehensiveness
• Topology
• Data Placement
• Message Routing

23
Open Problems Security

• Availability(copying with DOS attacks)
• Authenticity(trusting)
• Anonymity(Who has what)
• Access Control

24
Authenticity

• a

• Title Origin of species
• Author Charles Darwin
• Date 1859
• Body in an island

25
More than Just file Integrity

• a

• Title Origin of species
• Author Charles Darwin
• Date 1859
• Body in an island
• Check Sum

26
More than Fetching One File
T Origin Y? ADarwin B?
T Origin Y1859 ADarwin
T Origin Y1859 ADarwin Babcd
T Origin Y1800 ADarwin
T Origin Y1859 ADarwin
27
Solutions

• Authenticity Function A(doc) T or F
• At expert sites, at all sites?
• Can use signature expert?sig(doc)?user
• Voting Based
• Authentic is what majority says
• Time Based
• E.g oldest version(available) is authentuic

28
Added Challenge Efficiency

• Example Current music sharing
• Everyone has authenticity function
• But downloading files is expensive
• Solution Track peer behavior

Good peer
Good peer
Bad peer
29
How to track peer Behavior

• Trust vector v1, v2, v3, v4
• a b c d
• Single value between 0 and 1?
• Pair of values
• total downloads, good downloads ?

30
Trust Operations
Update?

• 1,.9,.5,0, 0

a
9
5
1,1,0,.3, 1
1,0,1,1,.2
b
c
3
1
2
3
d
e
31
Issues

• Trust computations in dynamic system
• Overloading good nodes
• Bad nodes can provide good content sometimes
• Bad nodes can build up reputation
• Bad nodes can form collectives

32
P2P Challenges

• Search?
• Resource Management
• Security Privacy ?

33
Resource Management
2
1
Capacity C2
Capacity C1
3
Capacity C3
34
Supporting Peer to Peer User Communities

• Julita Vassileva

35
Outline of the paper

• Design a new system based on the experience
• Motivation for group formation
• User Modeling
• Global Behavior of the system
• Discussion

36
Subject of the Paper

• Design of Peer to Peer System
• Which kind of system?
• Support file end services sharing
• For Whom?
• For the Researchers and Group of learners

37
Assumptions

• Users have long term interests
• Likely to search repeatedly in the same area
• Topology of the network doesnt change

38
Comutella and Goals

• A P2P (Gnutella based) system for file and
  service sharing ? users share academic papers,
  help and advise
  
• Increase the performance
• Increase satisfaction of User
• Provide fast routing

39
I-Help a community of peers
..
Beca.
40
Routing Strategies

• FreeNet Moving data in the network and storing
  it at certain nodes
  
• Highest hit Select the neighbor that has the
  highest number of query related files
  
• User Interest
• Exchange reputation information
• Self Organizing Groups

41
User Interest Group

• Group users based on their interests
• User groups can overlap.
• Servers use compatible ontologies
• User strength based on number of hits and time
  that he last search the area
  

42
User Interest Groups
43
User Resources

• New created resource should be indexed by user
• Each Resources is associated with 2 measure of
  importance
  
• Subjective(Respect to user)
• Objective (Respect to the community)

44
User Relationship

• Whom user is download file
• Who download file from the user.
• Strength
• Balance
• Time

Balance
Time
Strength
User satisfaction
Update on relationship
Direction of Service
45
User Model

• Model of Users Interest
• Model of Users Resource
• Model of Users Relationship

46
Updating User Model
Update User Model
47
Modeling User Interest
Programming Languages
48
(No Transcript)
49
Modeling User Resources

• What Files are downloaded
• What User does with the downloaded file
• What services are downloaded/used by others
• User add/delete/access
  Subjective Importance
  
• Other Users request/Download/Use
  Objective Importance

Ranking
50
Modeling User Relationship

• Which servents user request service, and Which
  services respond to user
  
• Who Download files form user, Success of the
  download
  

51
User Relationships
Vesile
Bach
Jane
Bach
Heart System
SS1S2
52
Balance of Relationship
BXY NX?Y NY?X

• Server ranks request ? Balance
• ? Importance
• Ranking of the Request Bandwidth
• Balance?Contribution Consumption
• Bxy? ?Help Maintaining User Contribution Model
• ? Constant

53
Learn About Other Users

• Direct Servents sends queries in a particular
  area to find out servents that have resources and
  enter them to the list of Relationships
  
• IndirectBy requesting from friend servers the
  list of their relationship in a particular area
  
• Indirect learning is not a dynamic way for P2P
• One interest group will have the same list
  of relationships

54
Strength of the new acquaintance

• Voting among the sources
• Averaging strength of the sources along the path
• By multiplying them
• Subjective Objective relationship
• Contextual Information

55
Levels of Participation

• Bring new files, give help
• Provide disk space / processor time
• Dispatch requests
• Stay on-line
• Use and quit

56
How to motivate participation?

• Why do people offer their time and resources? do
  p Different people have different
  motivations
  
• Some are altruists
  (Altruistic)
  
• Some would help their friends and hope to make
  new friends through helping
  
• Some seek glory
• Some seek high marks
• or Money
• l

(Socially Motivated)
(Materialistic)
57
Motivating participation Motivating Participa
tion
Altruistic Users
58
(No Transcript)
59

Motivating Participation
Social ranking

• B

Visualizing the community

• Users as stars in the night sky
• Proximity indicates closeness of interests among
  users
  
• Brightness -- their importance for the community

60
Motivating participation Introducing an economy

• Services have costs (time, effort)!

• How to cash the money?
• Depends on the values of the community
• Real money
• Marks in a classroom setting
• In better Quality of Service!!
• Faster access to files/services
• Better files/services

61
Servent Specialization

• Servents can be specialized in two areas
• Resources
• Relationship Hubs\
• Decision Which group to specialize?
• Strong relationship? Not specialize
• So many queries? Specialize
• Motivation for the Server to keep Resources
• Need of resources herself
• Rewards(Qos)
• Motivation for relationship hubs to specialize
• Charge Minimum connection fee for each query that
  have hit
  

62
Conclusions and future

• Motivating participation a key in P2P service
  sharing communities
  
• There are various motivation strategies
• Free riders are not evil but not enough to
  construct P2P community
  
• User modelling is needed individual and social
• Future work
• to evaluate COMTELLA
• to integrate previously developed bargaining
  mechanisms and simulate the economy
  
• to investigate trust issues and servent
  specialization
  

63
A Decentralized Algorithm for Coordinating
Independent Peers An initial Examination

• Girish Suryanarayana
• and
• Richard Taylor

64
Outline

• Purpose of the paper
• K-server Algorithm
• Emergency Levels
• Communication between firefighters
• Evaluation of the algorithm
• Future work
• Conclusion

65
Purpose of the paper

• Testing the applicability of the peer-to-peer
  approach at the emergency situations
  
• Peer to peer infrastructure prototype is
  developed
  
• Approach of applying P2P techniques to the
  emergency response applications
  
• Study is used as a simulation to study the
  algorithm

66
Emergency Response Domain

• Most important characteristic
• Communication
• Completeness
• Accuracy of information
• Decision Making
• Drawbacks of the Centralized System
• Bottleneck
• Disastrous delays
• P2P Decentralized System
• Reduced Setup Time
• Reduce Resource Prize

67
K-Server Algorithm
Request
2
4
1
5
5
Energy
5
8
4
6
6
5
3
3
7
Firefighter
Metric Space
68
Firefighter

• Has PDA(communicationinfo storagecomputing)
• Important properties of firefighters
• Position(x,y)
• Max energy capacity
• Energy level
• Current task (resting, fighting fires)

69
Energy Level

• Energy Thresholds
• Critical Energy(1 of max energy)
• Minimum Energy(20)
• Maximum Energy(80)

70
Communication

• Radio Wave and proximity
• Strength of the signal is inversely proportional
  with the distance
  
• Message being received is directly proportional
  with the Strength of the signal

Radio Range
Not Radio Range
Proxy
71
Types of the messages

• AssistRequest
• ResourceStatus
• AbortMove
• FireOut
• Every message has a unique identifier and Hop
  count
  
• Hop count 0 rebroadcast message

72
Algorithm for the firefighters
AbortMove
Hop n
ResourceStatus
ResourceStatus
AssistRequest
ResourceStatus
ResourceStatus
AssistRequest
AssistRequest
AssistRequest
AssistRequest
FireOut
Hop n-1
AssistRequest
FireOut
FireOut
73
Evaluation of the Algorithm

• For the application hop count 5 is the ideal(if
  too high flooding)
  
• firefighters 1/ time extinguish fire
• Changing Threshold Energy effect time to
  extinguish fire
  
• Problem of finding fires
• While moving towards other fires might become
  farther than the others
  

74
Future Work

• Inefficient Utilization of the bandwidth?Evolution
  of routing protocols
  
• Simple Leader election algorithm?developed leader
  election algorithm
  
• Fires do not increase size and intensity?Firefight
  ers will be leave or added depend on the
  intensity and the size of fire
  
• Finding different wandering algorithms to
  discover fires more effectively
  
• Increase number of message and prioritize them

75
Conclusion

• Demonstrate that P2P solution can be applied to
  an emergency response situations effectively
  
• Developed novel variant distributed k-server
  algorithm
  
• All fires can be extinguished in the case of
  adequate number of firefighters
  
• Constructed P2P framework is scalable

76
Referances

• http//www-db.stanford.edu/peers/
• 1 D. Milojicic, V. Kalogeraki, R. Lukose et
  al., Peer-to-Peer Computing, Technical Report
  HPL-2002-57, HP Labs, 2002.
  
• G. Suryanarayana, R.Taylor A Decentralized
  Algorithm for Coordinating Independent Peers An
  Initial Examination, 2002
  
• J. Vassileva Supporting Peer To Peer User
  Communities,2002
  
• Gnutella.com http//www.gunutella.com