Docstoc

Business Models of P2P Companies

Document Sample
Business Models of P2P Companies Powered By Docstoc
					Business Models of P2P Companies

An outlook of P2P architecture usage in business today




                    Summer Semester 2003


                         Saad N. Ahmad
             saad.nasir.ahmad@student.hu-berlin.de




                    Humboldt University Berlin
          Faculty of Economics and Management Sciences
Abstract
The IT revolution has enabled common man to utilize the Internet resources extensively. The
existing commonly used client/server architecture of the Internet has facilitated countless
sources of information, exchanges and services in the cyberspace. The natural spread of these
resources on the Internet, as well as emerging issues like scalability, authority and autonomy
etc. are leading to the development of new alternative architectures for organizing information
and effecting interaction.

In this paper, we evaluate the ‘Peer to Peer’ (P2P) architecture and assess its application to the
business world today. In particular, we identify the most important relevant attributes of the
P2P model and look at selected implementations, which completely or partially use the
architecture. The P2P model is still in its development phase, with many questions
unanswered and a number of new emerging concerns. However, it is expected that things will
change rapidly in future, as the technology develops further and incorporates more innovative
ideas.




                                                2
Table of Contents

  Abstract ................................................................................................................................. 2
  Introduction .......................................................................................................................... 4
  The P2P Networked Computing Model ............................................................................. 5
  P2P Applications .................................................................................................................. 6
    Parallel Processing ............................................................................................................. 6
    Content Exchange and File Management........................................................................... 7
    Collaboration...................................................................................................................... 7
  Characteristics of the P2P Architecture............................................................................. 7
    Cost sharing........................................................................................................................ 7
    Higher reliability and scalability ........................................................................................ 7
    Privacy and Anonymity...................................................................................................... 8
    Resource Aggregation and Synergies ................................................................................ 8
    Greater Autonomy.............................................................................................................. 8
    Dynamic Environment ....................................................................................................... 8
    Ad hoc Collaboration ......................................................................................................... 8
  P2P and Business Models .................................................................................................... 8
  Current P2P based Business Projects................................................................................. 9
    AVAKI:.............................................................................................................................. 9
    SETI (Search for Extraterrestrial Intelligence) ................................................................ 10
    Groove.............................................................................................................................. 10
    Magi ................................................................................................................................. 11
    FreeNet ............................................................................................................................. 12
    Gnutella ............................................................................................................................ 13
    JXTA ................................................................................................................................ 13
  Conclusions ......................................................................................................................... 15
  References ........................................................................................................................... 16



List of Figures:
Figure 1. A simple comparison of P2P and Client/Server architecture……………………6


Abbreviations:
CPU: Central Processing Unit
DNS: Directory Naming Service
P2P: Peer-to-Peer
PC: Personal Computer
PDA: Personal Data Assistant
TTL: Time To Live




                                                                     3
Introduction
One of the direct consequences of the digital revolution of last century has been the
availability of information. At no previous time in the history of mankind was access to
information easier and more cost effective. The spread of networked computing has brought
countless benefits to the areas of academics, businesses, arts, sciences and the life of common
man in general.

However, with the massive spread of information technology and the ever-increasing number
of its users, a number of new concerns have been emerging. People in both academics and
practice have been questioning the emerging nature and underlying model of networked
computing. The widely accepted and implemented model of Client / Server has worked well
in the past in satisfying the needs of its users. However, both the scalability issues coupled
with ever increasing computing power, connection bandwidth, reliability and ever decreasing
costs have provided the users the luxury of even higher expectations. But surely enough, there
are more than one-way of doing things in the information networks. The quest for better has
resulted in development of alternative architectures for network computing with their own
special characteristics.

One of the lately emerging computing architectures has been the Peer-to-Peer model (P2P),
which has attracted the attention of researchers, developers as well as investors. The P2P
model can provide better scalability, fault tolerance, lower costs, more power and more
efficient utilization of resources. Consequently, the enthusiasm for the P2P model ranges from
hype to disbelief in its potential. At present the implementations and utilization of P2P
technique has not been perfected, and a number of areas are still in the development phase.
Whether P2P will catch-up with the Client/Server model remains to be seen.




                                              4
The P2P Networked Computing Model
Since Peer-to-Peer (P2P) computing is still in its developing phase, there is no one standard
definition for it. There are a number of definitions used by the P2P community.

Clay Shirky of O’Reilly and Associate define P2P as a class of applications that takes
advantage of resources – storage, cycles, content, human presence – available at the edges of
the Internet. Because accessing these decentralized resources means operating in an
environment of unstable connectivity and unpredictable IP addresses, P2P nodes must operate
out-side the DNS system and have significant or total autonomy from central servers [Shirky,
2001]

The Intel P2P working group defines P2P as the sharing of computer resources and services
by direct exchange between systems [p2pwg, 2003]

Alex Weytsel of Aberdeen defines P2P as the use of devices on the Internet periphery in a
non-client capacity [Veytsel, 2001]

Ross Lee Graham defines P2P through three key requirements:
a) They have an operational computer of server quality
b) They have an addressing system independent of DNS, and
c) They are able to cope with variable connectivity [Graham, 2001]

According to Dejan Milojicic, the term peer-to-pee refers to a class of systems and
applications that employ distributed resources to perform a critical function in a decentralized
manner. The re-sources encompass computing power, data (storage and content), network
bandwidth, and presence (computers, human, and other resources). The critical function can
be distributed computing, data/content sharing, communication and collaboration, or platform
services. Decentralization may apply to algorithms, data, and metadata, or to all of them. This
does not preclude retaining centralization in some parts of the systems and applications if it
meets their requirements. Typical P2P systems reside on the edge of the Internet or in ad-hoc
networks [Milojicic et al., 2002]

The P2P architecture allows anonymous peers to share their resources, with or without a
limited interaction with a centralized server. In its purest form there is no concept of server
and all participants are ‘equal’ peers. A peer gives some resources, and in return obtains other
resources (including information, processing resources, presence, request forwarding etc.) that
are essential for the functioning of the system and benefit all peers. In a way the P2P peers
play the role of client and server at the same time.

The peers have to live with limited and unreliable connectivity, a possible independent
addressing system and be able to share the role of the server [Ozzie, 2000]. Peers are
autonomous when other peers or a central authority does not wholly control them. Because of
this autonomy, peers cannot necessarily trust each other or totally rely on the behavior of
other peers. Scale and redundancy are of prime importance in the P2P architecture. Also
security and accountability are of some concern in the P2P architecture.

One of the underlying concepts behind P2P is increasing the decentralization of systems. As
the Internet connectivity grows and Internet based services increase, it becomes undesirable
and unpractical to entrust the system functioning with centralized monopolized and
administratively managed infrastructures. In addition, P2P provides a means to harness the

                                               5
power of vast amount of computing resources including storage, CPU cycles and connectivity
from the countless and increasing number of personal computers distributed around the globe.
Another engine behind P2P architecture is the benefits of implementing horizontal
technologies such as distributed computing, collaborative and content sharing software, file
sharing, messaging software etc. P2P architecture seems to be more suitable for such
applications.

P2P model became popular with the rapid spread of music exchange software ‘Napster’ and
later with its lawsuit over digital rights management by music companies. It is increasingly
finding applications in other areas like grid computing. Many industry giants like Sony, Intel,
HP, Sun and a number of other companies are supporting a P2P working group as well as
some of the new software development platforms incorporate P2P capabilities, like JXTA
from Sun and .Net technologies from Microsoft. In addition a lot of projects are in progress
in various academic institutions such as Chord [Stoica et al., 2001], Ocean-Store [Kubiatowitz
et al., 2000], PAST [Druschel and Rowstron, 2001], CAN [Ratnasamy, 2001], and Free Net
[Clark, 1999].




            Figure 1. A simple comparison of P2P and Client/Server architecture
                               Source: [Milojicic et al., 2002]

A P2P system can be pure or hybrid. A pure system is self organizing and does not need to
interact with any centralized server, while in a hybrid system a central server my be deployed
to obtain meta information (peer directory) or security credentials when a new peer joins the
system. Afterwards direct interaction between peers takes place. Gnutella and Free net are
examples of pure P2P architecture. Napster, Groove, Aimster, Magi, Imesh and Softwax are
all based on a hybrid P2P architecture. Lately, some intermediate architecture have also
emerged (KaZaa, etc.) whereby the super peers contain some information about resources
which is used by joining peers if they are unable to find the resources themselves.

P2P Applications
The P2P environment includes Internet, intranets and other ad hoc networks. The applications
are designed to run on all sort of computing platforms including PCs, handhelds, and
notebooks etc. The P2P applications can be roughly divided into the following three
categories: [Milojicic et al., 2002]

Parallel Processing
These Applications divide a large task into smaller ones that can be executed in parallel on a
number of independent computers (peer nodes). The underlying concept is to use the idle
computer cycles available on the infinite number of nodes (PCs) on the Internet, to execute
computing intensive jobs. A lot of the implemented applications execute the same job on
different machines, using different parameters (grid computing). Examples include
SETI@Home, demographic analysis, code breaking, risk hedge calculations, portfolio pricing,
market and credit evaluations etc. These applications however, have not been widely accepted
to belong to the P2P class by the computing community. Other applications execute different

                                              6
components on each peer node. Examples include Workflow, Java Beans and Web services
etc. [Milojicic et al., 2002]

Content Exchange and File Management
These applications store and retrieve information from various peers in the network. The
current implementations do not focus on reliability and leave it to the user to make intelligent
choices about file availability and retry downloads in case of failure. Examples of these
applications are Napster [Napster, 2003] and Gnutella [Gnutella, 2003], which allow peers to
make available and download (music) files. Some academic projects have also attempted to
optimize the P2P file systems [Ratnasamy et al, 2001, Bolosky et al, 2000, Kubiatowicz et al,
2000, Rowstron and Druschel, 2001, Gribble et al, 2001, Stoica et al, 2001]. In addition, new
search techniques have also been implemented collaborative filtering and mining techniques,
which build searchable indices over P2P networks. Examples include OpenCOLA
[OpenCOLA, 2003] and JXTA Search [Waterhouse et al,. 2002].

Collaboration
These applications allow real time collection and relay of information in a pure P2P model
without interaction with a central server. Instant messaging belongs to this class with
applications like Jabber, Yahoo and AOL instant messaging. Other emerging software
includes shared applications that allow peers to interact while working on the same
information simultaneously. Examples include Buzzpad [www.buzzpad.com, 29.1.03] and
distributed PowerPoint [Rice and Mahon, 2000]. Some collaborative game applications also
belong to this class like NetZ 1.0 by Quazal [www.quazal.com, 29.1.03], Scour Exchange by
CenterSpan, Descent [www.planetdescent.com, 29.1.03], and Cybiko [www.cybiko.com,
29.1.03]

Characteristics of the P2P Architecture
The P2P model operates at the edges of the Internet i.e. between the end user nodes. The
number of peer machines in the network effects the system performance as well as the
scalability i.e. the ability of an infinite number of new peers to join in the system. Due to the
ad hock nature of P2P environment, most of the applications are based on best effort
algorithms, which may or might not be sufficient for certain (business) applications. There are
a number of unique characteristics of the P2P model that effect the design of applications for
this architecture:

Cost sharing
In the client/server model, it is the server side that bears most of the costs. P2P model can
spread the cost among the peers by using the existing infrastructure and distributing and
eliminating the running costs. In addition, the aggregation and utilization of otherwise unused
resources results in reduced net marginal costs. However, it is also important that the costs are
shared reasonably equally among the autonomous peers. [Milojicic et al., 2002]

Higher reliability and scalability
System scalability and reliability are important in an autonomous P2P environment without
the presence of any central authority to take care of these essential parameters. To this end,
new algorithms have been developed for search and resource discovery for both the existing
systems [REFS], as well as new P2P development platforms (CAN [Ratnasamy, 2001], Chord
[Stoica, 2001] and PAST [Rowstron, 2001]).



                                               7
Privacy and Anonymity
In a P2P model, all activities are local to the peers without having any central server, which
allows the peers a greater degree of autonomy and control over their data and resources. The
P2P system design allows the peers to avoid providing any user information. One example is
Freenet, which uses message forwarding as well as probabilistic algorithms to hide the
requester of a service. [Milojicic et al., 2002]

Resource Aggregation and Synergies
In P2P model, every node (peer) contains certain resources, which can be aggregated and used
by special applications. Aggregation can leads to synergies that increase the scope of
utilization of the whole system. CPU intensive simulations (SETI@Home, Distribute.net,
Endaevours) can aggregate computing power to provide a virtual super computer. Distributed
file systems (Napster, Gnutella, Freenet etc.) use aggregation to increase disk space for
storing community data as well as bandwidth for moving the data around. Aggregation calls
for low cost interoperability of diverse resources. [Milojicic et al., 2002]

Greater Autonomy
P2P model allows grater autonomy for its peers by eliminating the need to rely on and follow
the rules set by a single central resource provider. This autonomy allowed the Napster users to
bypass the licensing restrictions on music files. Individual peers running their own mini
servers (PCs) could offer their files without having to account for it. As the PC power
increases, the peers could offer more and more services. [Milojicic et al., 2002]

Dynamic Environment
P2P model is based on a highly dynamic environment, whereby the peers (resources) join and
leave the system continuously. Therefore the underlying applications need to incorporate
special features to ensure that if a peer leaves the system in the middle of an operation, an
alternate resource is available to complete the job. Nothing is irreplaceable in a true P2P
environment. [Milojicic et al., 2002]

Ad hoc Collaboration
The dynamic nature of P2P model, leads to an environment whereby the peers join and leave
based on their location and interests. P2P applications take into consideration changes in the
group of participants, and incorporate communication and collaboration on ad hoc basis. P2P
systems therefore build their own self-organizing structures at the spur of the moment,
depending on who and what is available. [Milojicic et al., 2002]

P2P and Business Models
Attempts have been made by a number of companies to utilize the P2P architecture for
making money. However, to date, the only successful business model is based on benevolent
users donating their CPU resources for scientific work (SETI@Home). Although, some
content sharing software client developers do provide functionality enhancements based on a
fee as well as display third party advertisements in the client console in order to finance the
client programming projects. A P2P model for mega scale business is yet to be developed.
Some of the main obstacles in this direction are concerned with secure payment system and
ratings. [Milojicic et al., 2002]

Inherent in the privacy and anonymity of P2P environment is the issue of lack of
responsibility. A lack of a central authority makes it difficult to enforce contracts. In addition,


                                                 8
there is no true P2P payment mechanism so far, which facilitates exchanges without
depending on some sort of intermediate authority and without infringing the privacy concept.
An individual rating system (like used on Ebay) can help solve the trust and contract
enforcement problem, while an electronic payment mechanism (incorporating a card reader
and e-cash) can be developed to solve the payment issues. In addition, some variant of escrow
may be applied to P2P architecture in order to ensure smoothness of transactions. [Milojicic et
al., 2002]

Current P2P based Business Projects
In this section we consider some projects that utilize P2P elements and analyze their business
models. This selection is only meant to give a flavor of the various attempts being made to
incorporate the benefits of P2P architecture in useful applications. A pure P2P based business
model does not exist so far for the reasons mentioned above.

AVAKI:
Website: www.avaki.com (29.9.03)

Function: Avaki takes a heterogeneous network of computing resources and provides a single
virtual computer view. It is marketed as a middleware platform for enterprise level
computing. The vision is to create a massive parallel computing environment for faster
execution of applications. Current work focuses on incorporating robust security, performance
management and failure detection features to simplify grid administration. In addition, work
is being done to make Avaki interoperate able with emerging P2P, distributed and pervasive
computing standards such as JXTA.

Business Model and Applications: Avaki markets its products as complete data and compute
grid solutions to enterprises. These are currently being evaluated at various research and
benchmarking labs. Avaki faces competition from other HPC vendors such as Platform
Computing. In addition, Avaki also provides high-end calculation intensive solutions for
science applications including biochemical simulations, DNA sequence comparison and 3D
structures space study.

History: Avaki started as a university research project initiated by Andrew Grimshaw at the
University of Virginia in 1993 (called Legion at that time) [Grimshaw et al., 1994]. Then the
idea was to achieve a unified ´virtual computer` view of the various computing resources
scattered around the US. In 1998 the project developed into a commercial venture called
Applied MetaComputing. In 2001 it was launched as Avaki Corp. Which today provides
enterprise level distributed computing solutions.

Design: Avaki is based on Object Oriented paradigm. It is built as a layered virtual machine
with scalable, distributed control. The application has three layers of services. The core
services perform the tasks required to map the system to a network of computing resources.
The system Management services allow Avaki system to monitor and control the local
resources. The Application Services enable the construction of high performance computing
and collaborating services. Hardware redundancy is built into the system; however fault
tolerance as well as automatic error detection and correction, vital for such a large systems are
proving to be a challenge. Avaki dynamically configures the grid and automatically migrates
failed jobs to different locations of the system. The user can also specify the desirable level of
fault tolerance. Security features are also incorporated into the system.


                                                9
Performance: Avaki is designed as a virtual machine that incurs an overhead. However, in
most of the cases it is more than offset by the performance gain through parallel processing.
The actual increase in speed of execution depends upon the nature of the applications used.

SETI (Search for Extraterrestrial Intelligence)
Website: http://setiathome.ssl.berkeley.edu/ (29.9.03)

Function: SETI@Home project analysis the radio emissions received from space via the
giant Arecibo telescope. It analyzes the signals using the processing power of millions of idle
PCs connected through the Internet. The idea is to search for extra terrestrial radio emissions
form possible intelligent civilizations situated outside our solar system.

Business Model and Applications: SETI@Home is a scientific project made possible by
government grants. Benevolent people in the name of science donate the computing power.
The project proved that distributed computing technology could be applied to real situations.
A pure commercial application of this approach is debatable at the moment.

History: SETI@Home is a scientific research project of the US government. It aims to build a
super computer by aggregating the power of countless idle computers connected via Internet.
The project enjoys wide acceptance and popularity worldwide, and uses the power of more
than three million computers to produce dozens of TeraFlops of processing power.

Design: The major components of SETI@Home are a database server and the clients. The
client software runs on several different operating systems and is available as a screen saver
module. The server dispatches jobs to each peer and collects the processed results. The
database is extremely scaleable and stable even with more than three million registered users.
Fault resilience is built into the system, since the batch jobs can be as long as 10 hours, and
interruptions due to machine shutdowns or user logons require seamless recovery. The system
relies on a check pointing system, whereby resume able dataset is saved on the server every
10 minutes, allowing the computation to restart from the last saved dataset in case of an
interruption. The resulting reliability more than offsets the overhead of this check pointing
system.

Performance: System possesses excellent horizontal scalability (number of users). A
bottleneck in vertical scalability architecture exists due to the presence of a single server for
coordination of all the operations. However, the system has been able to take the load of a
huge number of users.

Groove
Website: http://www.groove.net/ (29.9.03)

Function: ’Groove workspace’ enables communication, content sharing and provides tools
for joint activities [Leigh and Benyola, 2001]. It is a collaborative P2P system, targeted at
users of Internet, intranets, mobile phones, PDAs as well as other mobile devices. The
underlying vision behind Groove is to allow users to communicate directly with other users
without relying on a server [Suthar and Ozzie, 2000]. Security, privacy and flexibility are the
key features of the product.

Business Model and Applications: Groove Networks plans to license its infrastructure
platform to corporations and third-party integrators [Rice and Mahon, 2001]. Groove enables
elimination of network administration costs, minimization of dependency on server
infrastructure and improves availability. The competitors in collaborative P2P market include
                                               10
Microsoft and Endaevour Technologies. Applications of Groove include, voice over IP,
instant messaging, chat and threaded discussions in communication areas. Content sharing
and collaboration applications include shared files, images and contacts as well as group
calendaring, group drawing and editing, and co-Web browsing.

History: Groove was founded in1997 by the developer of Lotus Notes, Ray Ozzie. The first
product version was released in 2001.

Design: Groove supports Internet, intranets and private networks. Groove users are
authenticated and secure key technologies are used to provide data privacy and integrity, both
on disk and wire. Private user data is not shared on the network, and users are allowed to
handle disconnections transparently. Groove supports a number of reusable components and
its components are signed. Security is implemented via public/private key management.
Storage is based on XML object store. Fault resilience is implemented by resending data to
failed nodes on rejoining. The groove layer works between the application logic and
command execution layers. [Groove Networks, 2003]

Performance: A number of companies have been successfully using the software. The
customers are expected to increase, as the software incorporates more new features.

Magi
Website: http://www.endeavors.com (29.9.03)

Function: Magi is a P2P infrastructure platform for building cross platform collaborative
applications. The vision behind Magi is enabling information sharing and messaging on any
existing-standards based web device (as opposed to proprietary interface and communications
standards). The high inter-operate ability as well as web based application design makes Magi
suitable for wide range of devices.

Business Model and Applications: The ‘Magi Enterprise’ application links the corporate
functions teams with the operations teams for sharing files, instance messaging and chat. It is
based on the workflow models used in collaborative business environments. The software
works on a number of devices including PDAs and pocket PCs. The application integrates
with standard software and can embed collaborative processes into enterprise wide
applications and B2B products.

History: Magi was originally a research project at the University of California, Irvine headed
by Greg Bocler. The software was initially funded by DARPA as the largest (at that time)
non-Sun java project in US. Endaevours Technology was founded in 1998, and later taken
over by Tadpole Technologies. The first Magi version was released in 2000 followed by the
Enterprise edition in 2001.

Design: Magi uses Web-based standards including HTTP1.1, WebDAV (for remote
authoring of documents), and SWAP/Wf-XML (for remote process monitoring and control) to
enable communications between the applications, joined over an enterprise network or the
Internet. The main components of Magi framework include a micro Apache HTTP server
linking every instance of Magi, a set of core services and a generic extensible interface. The
entire infrastructure is in java with web interface via applets, which requires a minimal web
server to be present and running. Each instance of Magi acts as both a client and server for
P2P communications. Fault resilience is build by guaranteeing message queuing for current
offline users, who receive the message once destined user comes on line. A single central
dynamic DNS is used as IP address directory, introducing a single point of failure as well as
                                              11
scaling issues. Similarly a single central Certification Authority also performs user
authentication.

Performance: Magi has performed well with a variety of device although some
interpretability issues have emerged from time to time.

FreeNet
Website: http://freenet.sourceforge.net/ (29.9.03)

Function: FreeNet is an anonymous file sharing system. Anonymity is of prime importance
in the system design, in that users are not allowed to know who placed the files or who is
requesting them. Even the operator of a Freenet node is not allowed to know about the
contents of the local shared disk space. The vision is to provide an absolute anonymous
method for storing and retrieving information. Freenet developers value non-censorship and
freedom over intellectual property rights.

Business Model and Applications: Freenet is a non-profit open source project. As such no
profit making business model has been envisioned. The systems main application is
information storage and retrieval. The anonymity and strong cryptographic properties makes
Freenet an ideal candidate network based backup system, however the probabilistic guarantee
of data retrieval makes it somehow not very reliable for this application.

History: Freenet was developed at the University of Edinburgh in 1999 by Ian Clarke [Clark
1999, Clark et al. 2001]. The Mike Godwin Public development of Open Source Freenet
began in 2000

Design: Freenet is available in an Open Source reference implementation with a well-defined
protocol code. It represents the pure form of P2P, since complete decentralization is
incorporated into the system. The basic storage unit is a file and sets of files are maintained on
every node up to the maximum space allocated. When the space is filled up, the files are
overwritten using the ‘least recently used’ strategy. The files are identified by keys using the
SHA-1 [SHA-1 1997] function. The public key becomes the file identifier, while the private
key is used to sign the file for file integrity check.

The basic allowed operations in Freenet are insertion and searching of files. Message
forwarding is used between the nodes to find a file or location for inserting a new file. The file
keys help in routing these messages. Freenet attempts to cluster files with similar keys on a
single node and creates routing tables to optimize searches. After the message reaches a node
with the required file, the whole file is returned and gets copied to all the intermediate nodes
before it reaches the requester. In this way popular files get highly replicated. Inserts are done
by first a routed search to located duplicates of file key, and then (if none exists) replicating
the file all the way up to its destination file cluster based on its key. New nodes announce
themselves by performing a search, which propagates through the network. However they
need to find at least one peer to start this process, for which Freenet offers some out-of -band
means like a limited web listing.

Performance:
Simulation studies have shown that the search route length grows logarithmically with the
number of users [Clarke et. al. 2001]. Other performance benchmarking is difficult because of
the highly anonymous and probabilistic nature of the application.


                                               12
Gnutella
Website: http://www.gnutella.com/ (29.9.03)

Function: Gnutella is a file sharing protocol, which allows users to search for and download
files, shared by others users on the Internet. The underlying vision is to provide a purely
distributed file sharing solution. Gnutella is open source, therefore users can implement the
protocol and share files with anonymity but also a certain degree of uncertainty.

Business Model and Applications: Gnutella is an Open Source protocol, free for download
and implementation by the interested. The main application of Gnutella has been sharing of
music files. A number of companies are using the protocol to develop enterprise software for
project management as well as academic communications software. Gnutella clients are also
coded by some companies and available as freeware but with third party advertisement boxes.

History: The Gnutella file sharing technology was developed and introduced in March 2000
by employees of AOL’s Nullsoft division. The program was shared via AOL website, but
following the Napster trail, it was seen as a possible threat for Warner Music and EMI. AOL
was also in the middle of merger talks with the record companies, therefore the program was
taken offline in just one day. But the damage has already been done, and the hackers who
produced clones working with the same protocol already cracked the Gnutrella. Afterwards,
versions of original Gnutella were seen to exchange files and work with the clones over
Gntella Network.

Design: Gnutella is simply a protocol that can search and share files among users. To start
with a user must know the IP address of at least one more Gnutella node (which can be
obtained from the Gnutella web site). The user can issue a query for a file to his known nodes,
which may respond or forward the request to their known Gnutella nodes. The query
propagation is controlled by a TTL (time to live) field. The scalability of protocol is
questionable in that the number of queries and responses increase exponently with each hop.
Gnutella does not provide fault resilience, and studies have shown [Adar and Huberman,
2000, Saroiu et al., 2002] that only a small fraction of Gnutella users actually remain online
long enough to respond to queries from other users. Users are therefore expected to repeat
their query or download on failure.

Performance: A number of companies have tried to overcome the limitations of the original
protocol. The most popular clones are Limewire [Limewire, 2003], ToadNode [ToadNode,
2003], and BearShare [BearShare, 2003]. Kotzen has studied the Gnutella usage and claims
that in late 2000 only 10% of download attempts were successful, but by March of 2001 the
number grew to over 25% [Kotzen, 2001]. In addition, the number of Gnutella users has
increased. However, Kotzen reports that a maximum of 11,000 users has been seen on the
network at any one time. While this represents an increase over previous, it does not provide
any proof of the scalability or performance of the Gnutella network for the targeted thousands
or millions of nodes.

JXTA
Website: http://www.jxta.org/ (29.9.03)

Function: The JXTA project provides an open innovative collaboration platform that
supports a wide range of distributed computing applications and enables them to run on any
digital device. The vision is to provide a general-purpose network programming and
computing infrastructure.


                                              13
Business Model and Applications: JXTA has been released as Open Source code, and many
developers are using it to build applications (e.g., event notification, file sharing, P2P email).
Sun is also working with P2P companies that are committed to using JXTA with their
technologies. Some of the features of the Platform are:

Interoperability: easy peer discovery, participation in community activities and seamlessly
offering services across different P2P systems and different communities
Platform independence: works independently of programming languages (C, Java etc.),
operating system (Windows, UNIX, Solaris etc.), and networking platforms (TCP/IP,
Bluetooth etc.)
Ubiquity: JXTA is implement able on every digital device including appliances, desktop
computers, and storage systems.

History: The JXTA project was started by Sun in April 2001 as a platform to develop a wide
range of distributed computing applications. Since then the number of JXTA users has been
developing rapidly. The JXTA shell was the first application providing a command line
interface to the underlying core applications.

Design: The JXTA project approaches the P2P space from the lower level as it proposes an
entirely new infrastructure with no direct relation to other, existing P2P systems (e.g.,
Gnutella and Napster). It provides the functionality in multiple layers with basic functions in
the core layer, higher services layer that expand the functions of the core and finally an
applications layer demonstrating the broad applicability of the platform. It has its own
distributed search capability and peer-group based communication, security, logical locality
and performance concepts. A peer can be a part of multiple groups at the same time. JXTA
provides core protocols for functions like peer discovery, peer group memberships and
monitoring. The data moves in the form of XML documents via asynchronous unidirectional
communications channels.

Performance: A first version of JXTA has been released for Linux, Solaris and Windows
operating systems. Many independent efforts are underway to build services and applications
with JXTA. Some scalability issues like unique global naming and ID still need to be
resolved.




                                               14
Conclusions
P2P model is emerging as an important alternative to the tradition client/server network
computing. The salient features of the P2P model include decentralization, lower costs,
anonymity, ad hoc behavior, scalability, fault resilience and self-organization. On the other
hand a lack of security, accountability and reliable payment system is preventing P2P model
from entering the mainstream business arena.

The existing P2P applications are based around the concepts of Parallel processing, content
exchange and file management as well as collaboration. A number of researchers,
programmers and investors are engaged in further developing the P2P architecture and
producing new applications. Many commercial companies already have running products that
utilize at least some of the useful features of P2P architecture. In addition some pure P2P
based applications also exist. We looked at a selection of P2P applications and found that
most of the present implementations try to benefit from some characteristics of the P2P model
and try to integrate the Client/Server model within the new P2P architecture. This hybrid
model has beinfits of both the original models.

P2P technology is still in its development phase and a successful and widely accepted
commercial business model does not exist at this point in time. Entrepreneurs are seeking
ways to move the technology toward this end. However, P2P still remains an experimental
and shareware technology. But lots of changes are expected to happen in future in this
promising area and we may yet again witness another technology revolution based on the P2P
model.




                                             15
References
ADAR, E.AND HUBERMAN, B., 2000. Free Riding on Gnutella. First Monday, vol 5, no 10
(October).
(see also www.firstmonday.dk/issues/issue5_10/adar.) (25.9.03)

BEARSHARE, 2003.
www.bearshare.com. (25.9.03)

BOLOSKY, W., DOUCEUR, J., ELY, D., AND THEIMER, M., 2000. Feasibility of a
Serverless Distributed File System Deployed on an Existing Set of Desktop PCs. Proceedings
of SIGMETRICS, Santa Clara, CA, USA, Jun, 2000.

CLARKE, I., 1999. A Distributed Decentralized Information Storage and Retrieval System,
unpublished      report,   Division     of     Informatics, University of    Edinburgh.
www.cs.cornell.edu/People/egs/615/freenet.pdf (20.9.03)

CLARKE, I., SANDBERG, O, WILEY, B., HONG, T.W., 2001. Freenet: A Distributed
Anonymous Information Storage and Retrieval System. Designing Privacy Enhancing
Technologies: International Workshop on Design Issues in Anonymity and Unobservability,
LNCS 2009, ed. by H. Federrath. Springer: New York.

DRUSCHEL P. AND ROWSTRON, A., 2001. PAST: A Large-Scale, Persistent Peer-to-Peer
Storage Utility, HotOS VIII, Schloss Elmau, Germany, May 2001.

GNUTELLA. 2003. The Gnutella home page
http://gnutella.wego.com/ (25.9.03)

GRAHAM, R.L. 2001. Peer-to-Peer: Toward a Definition, Lecture. Proceedings of IEEE:
International Conference on Peer to Peer Computing 2001, Sweden, 26-27August 2001

GRIBBLE, S., HALEVY, A., IVES, Z., RODRIG, M., AND SUCIU, D., 2001. What Can
Peer-to-Peer Do for Databases and Vice Versa? Proceedings of the WebDB: Workshop on
Databases and the Web, Santa Barbara, CA, USA.

GRIMSHAW, A.S., WULF, W.A., FRENCH, J.C., WEAVER, A.C., REYNOLDS JR., P. F.,
1994. Legion: The Next Logical Step Toward a Nation wide Virtual Computer. UVa CS
Technical Report CS-94-21, June 8, 1994.

GROOVE NETWORKS. 2003. Business Solution: Product Design. Groove Networks White
Paper.
www.groove.net/pdf/scen-design.pdf (20.9.03)

GROOVE NETWORKS, 2003. Microsoft Visual Studio .Net and Groove Workspace. Brief
Groove Networks Brief.
www.groove.net/pdf/ms_vsnet.pdf (25.9.03)

KOTZEN, M., 2001. Dramatic Improvements in the Gnutella Network Since Late 2000.
www.limewire.com/index.jsp/net_improvements (20.8.03)



                                           16
KUBIATOWICZ, J., BINDEL, D., CHEN, Y., CZERWINSKI, S., EATON, P., GEELS, D.,
GUMMADI, R., RHEA, R., WEATHERSPOON, H., WEIMER, W., WELLS, C., AND
ZHAO, B., 2000. OceanStore: An Architecture for Global Scale Persistent Storage.
Proceedings of the Ninth international Conference on Architectural Support for
Programming Languages and Operating Systems (ASPLOS 2000), November 2000.

LIMEWIRE, 2003
www.limewire.com. (25.9.03)

LEIGH, P.AND BENYOLA, P., 2001. Future Developments in Peer Networking. Equity
Research, WhitePaper, Raymond James & Associates, INC.

MILOJICIC, D.S., KALOGERAKI, V., LUKOSE, R., NAGARAJA, K., PRUYNE, J.,
RICHARD, B., ROLLINS, S., XU, Z., 2002. Peer to Peer Computing. HP Laboratories Palo
Alto, HPL-2002-57. March 8th 2002

NAPSTER, 2003. The Napster home page,
www.napster.com (25.9.03)

OPENCOLA, 2003.
www.opencola.com (25.9.03)

OZZIE, J., 2000. SYSTEMS INTEGRATION WITH GROOVE. Groove Networks
Presentation.
www.groove.net/developers/Presentations/GrooveSystemsIntegratoin.ppt (20.9.03)

PEER-TO-PEER WORKING GROUP. 2003.
http://p2p.internet2.edu/ (25.9.03)

RATNASAMY, S., FRANCIS, P., HANDLEY, M., KARP, R., SHENKER, S., 2001. A
Scalable Content Addressable Network. Proceedings of the SIGCOMM, pp 161-172.

ROWSTRON, A.AND DRUSCHEL, P., 2001. Storage Management and Caching in PAST, a
Large-Scale, Persistent, Peer-to-Peer Storage Utility. Proceedings of SOSP, pp 188-201

SAROIU, S., GUMMADI, P., AND GRIBBLE, S., 2002.A Measurement Study of Peer-to-
Peer File Sharing Systems. MMCN, San Jose, CA, USA January 2002.

SHA-1. 1997. American National Standards Institute, American National Standard X9.30.2-
1997: Public Key Cryptography for the Financial Services Industry Part 2: The Secure Hash
Algorithm (SHA-1).

SHIRKY, C., 2001.WhatisP2P...and what Isn’t. An article published on O’Reilly Network.
www.openp2p.com/lpt/a//p2p/2000/11/24/shirky1-whatisp2p.html (25.9.03)

STOICA, I., MORRIS, R., KARGER, D., KAASHOEK, F., AND BALAKRISHNAN, H.,
2001. Chord: A Scalable Peer-to-peer Lookup Service for Internet Applications. Proceedings
of the SIGCOMM, pp 149-160.

SUTHAR, P.AND OZZIE, J., 2000. The Groove Platform Architecture. Groove Networks
Presentation.

                                           17
www.groove.net/developers/Presentations/GrooveApplicationArchitecture.ppt?tip=Ic
(20.9.03)

TOADNODE, 2003.
www.toadnode.com. ( 25.9.03)

VEYTSEL, A., 2001. There is no P-to-P Market... But There is a Market for P-to-P. Aberdeen
Group Presentation at the P2PWG, May 2001.

WATERHOUSE, S., DOOLIN, D.M., KAN G., FAYBISHEN-KO, Y., 2002. Distributed
Search in P2P Networks. IEEE Internet Computing 6(1): 68-72. January-February.




                                           18

				
DOCUMENT INFO
Shared By:
Tags:
Stats:
views:25
posted:8/1/2011
language:English
pages:18
Description: The IT revolution has enabled common man to utilize the Internet resources extensively. The existing commonly used client/server architecture of the Internet has facilitated countless sources of information, exchanges and services in the cyberspace. The natural spread of these resources on the Internet, as well as emerging issues like scalability, authority and autonomy etc. are leading to the development of new alternative architectures for organizing information and effecting interaction.
cao yunding cao yunding http://
About