Wednesday, April 18, 2007

P2P: From Internet Scourge to Savior

[Here is a set of pointers from various sources on the growing phenomena of P2P applications. Many network operators and universities are trying to limit the impact of p2p applications, as only a small number of P2P users can cause significant traffic congestion on today's networks. But to my mind this is an opportunity, rather than problem: the increasing computational and storage capability at the edge means we may have to fundamentally rethink the architecture and business models of the current Internet.--BSA]

P2P, now for Pretty Much Everything

[Excerpts from Om Malik's column]

At the dawn of the broadband era, peer-to-peer technology became closely associated with music file sharing, thanks to programs like Napster and Kazaa. Later, the emergence of protocols such as BitTorrent linked P2P to movie and television downloads. Over the years, P2P has found many legitimate uses and has found a way into everyday life. Why… even Akamai, which had scoffed at P2P, decided to acquire Red Swoosh.

RawFlow is a 6-year-old company that has developed a peer-to-peer streaming software platform for sharing live video streams. It is currently developing a personal broadcasting technology called Selfcast, also based on P2P technologies.

Nevertheless, the Akamai-Red Swoosh deal, announced last week, prompted me to think about how pervasive P2P really has become in our lives.

* P2P telephony: Skypeand its $4.2 billion price tag - that says it all.
* P2P TV: Joost and Babelgum are just a start. More like Zattoo are joining the party.
* Personal P2P: P2P sharing of photos, videos and other files with family and friends is becoming increasingly common place. The list of start-ups chasing this nascent market is growing by the month
* P2P Video Delivery is growing in popularity, especially in places where 10 megabit/s broadband connections are commonplace.
* P2P data syncing between computers.
* Distributed computing is another area where we have seen P2P technology shine. SETI@Home is a good example.

[From a posting on Dewayne Hendrick's list -- BSA]

New peer-to-peer video-downloading options should please Internet
users--and stave off a network meltdown.
By Wade Roush

Every few years, someone predicts the imminent collapse of the
Internet. Bob Metcalfe--Ethernet inventor, 3Com founder, and
Technology Review patron--famously said at a 1995 Web conference that
he would eat his words from a pessimistic Infoworld column if the Web
didn't disappear within a year under the strain of traffic overloads
and other problems. In April 1997, Metcalfe contritely drank a
milkshake containing the torn-up bits of his column. In 2004,
Helsinki University of Technology professor Hannu Kari said spam and
viruses would kill off the Internet by 2006. The fact that you're
looking at this website now means Kari was wrong.

The point is that even the sharpest innovators and entrepreneurs have
often had a hard time seeing how the next hurdle in the Internet's
growth would be overcome--only to be surprised by some new resource,
technology, or business idea that emerges in the nick of time. (In
both 1996-97 and 2005-2006, Internet service providers responded in
part by adding more bandwidth to their networks.)

But will that pattern hold out forever? The dam-breaking success of
YouTube and Apple's iTunes Video Store--neither of which existed
prior to 2005--has unleashed a huge new flow of digital video on the
Internet, and many consumers now spend hours a day streaming or
downloading everything from home movies to live sports and prime-time
TV series. Because video files are so large compared with the Web
pages and e-mail messages that used to dominate Internet traffic,
backbone lines are under strain, and backbone operators such as AT&T
and Verizon and Internet service providers such as Comcast are facing
new costs they can't easily recoup, given the flat-rate pricing of
most consumer broadband Internet access plans.

Hui Zhang, a computer scientist at Carnegie Mellon University who
studies broadband networks, says that "2006 will be remembered as the
year of Internet video. Consumers have shown that they basically want
unlimited access to the content owners' video. But what if the entire
Internet gets swamped in video traffic?"

This time around, the Internet may be saved by the unlikeliest of
rescuers: the builders of peer-to-peer file-sharing networks. In the
minds of many consumers--and many studio executives--P2P networks are
still synonymous with digital piracy. After all, Napster, Kazaa, and
other early peer-to-peer networks were playgrounds for copyright
violators, who downloaded millions of music files they hadn't paid
for. But today a number of researchers and entrepreneurs are arguing
that peer-to-peer technology--which allows network members to
retrieve content by tapping into the hard drives of other members who
have already downloaded that content--is also great for distributing
legitimately purchased, copyright-protected music and video. It might
even lessen the burden on service providers and content distributors.


[From a posting on Dewayne Hendrick's list -- BSA]
[Note: This item comes from reader Jock Gill. DLH]
Subject: Bootstrapping the Long Tail in Peer to Peer Systems

You might find this paper of interest. Jock

Bootstrapping the Long Tail in Peer to Peer Systems

Bernardo A. Huberman, HP Labs, Palo Alto, CA 94304
and Fang Wu, HP Labs, Palo Alto, CA 94304

1 Introduction
The provision of digitized content on-demand to millions of users
Presents a formidable challenge. With an ever increasing number of fixed and
Mobile devices with video capabilities, and a growing consumer base with
different preferences, there is a need for a scalable and adaptive way of
delivering a diverse set of files in real time to a worldwide consumer base. Providing such varied content presents two problems. First, files should be accessible in such a way that the constraints posed by bandwidth
and the diversity of demand is met without having to resort to client server
architectures and specialized network protocols. Second, as new content is
created, the system ought to be able to swiftly respond to new demand on specific content, regardless of its popularity. This is a hard constraint on any distributed system, since providers with a finite amount of memory and bandwidth will tend to offer the most popular content, as is the case today with many peer-to-peer systems.

The first problem is naturally solved by peer to peer networks, where
Each peer can be both a consumer and provider of the service. Peer to peer
net-works, unlike client server architectures, automatically scale in
size as demand fluctuates, as well as being able to adapt to system failures.
Examples of such systems are Bittorrent [4] and Kazaa, who account for a sizable percentage of all the use of the Internet. Furthermore, new services like the BBC IMP,( show that it is possible to make media con- tent available through a peer-to-peer system while respecting digital rights. It is the second problem, that of an adaptable and efficient system
Capable of delivering any file, regardless of its popularity, that we now
solve. We do so by creating an implementable incentive mechanism that ensures the existence of a diverse set of offerings which is in equilibrium with the available supply and demand, regardless of content and size. Moreover, the mechanism is such that it automatically generates the long tail of offerings which has been shown
it only takes about 10 BitTorrent users bartering files on a node (of around 500) of traditional shared IP network to double the delays experienced by everybody else.