2001 P2P Networking Overview -- Executive Summary

2001 P2P Networking Overview

The Emergent P2P Platform of Presence, Identity, and Edge Resources

By Clay Shirky, Kelly Truelov, Rael Dornfest & Lucas Gonze
September 2001 (est.)
ORP2PREP, Order Number: ORP2PREP
312 pages (est.), $495.00

Executive Summary

Some have called peer-to-peer computing the third generation of the Internet. The first generation, so the reasoning goes, was the "raw" Internet: indispensable for professional computer users and other early adopters, but irrelevant to the rest of the world. The second generation was the Web, which introduced the Internet to the masses. This third generation makes new services available to end users cheaply and quickly by making use of their PCs as active participants in computing processes rather than just Web browsers. More to the point, P2P is doing this in radical ways that upend traditional means of doing business. That's why it's a disruptive technology.

That's the buzz, anyway. P2P is a very large umbrella, and in the middle of 2001 many companies are scrambling to huddle under it. Some of those companies deserve the P2P moniker more than others, and some companies are, quite frankly, poorly conceived. Still others have good products, but are unlikely to ever be very profitable. In this report, we'll explore the companies, their products, the technologies behind them and their prospects for success.

P2P Overview

P2P is a mindset, not a particular technology or an industry.

P2P provides a new way of utilizing distributed resources, which we say are found at the edge of network. Since typical networks consist of centralized servers, many people call this mindset "decentralization," and that's approximately right.

To fully grasp P2P, you must think in terms of "PIE:" Presence, Identity and Edge Resources.

Working backward, Edge Resources include content, storage, cycles, bandwidth and even human attention, accessible at the edges of the network, which is to say resources accessible on individual PCs. Identity is simply a name for one of those resources -- a machine, a document or a person. Presence is the ability to detect whether that resource is accessible in real time.

P2P architectures offer powerful approaches for fundamental problems of the Web.

The Web has some fundamental architectural flaws, principally related to the over-reliance on giant server farms in the center of the network, and the under-reliance on the untapped resources on individual clients at the edges. P2P architectures, by tapping these underused resources, offer powerful approaches for solving these problems, problems such as bandwidth cost, denial of service attacks, cost of maintaining 24/7 systems, etc.

P2P means being able to choose the appropriate balance between centralization and decentralization.

P2P does not present a binary choice between centralization and decentralization, but represents a way of decentralizing those aspects of a system that can be better handled at the edges of the network. Most P2P systems are in fact "impure," relying on a central server or one or more "super peers" to bootstrap connectivity or resource identification. It's rare that a purely decentralized system is the right design choice anyway. Napster and other key P2P applications centralize specific functionality. The press coverage of P2P - both positive and negative - often gets this wrong, painting P2P as 100 percent decentralization, with all peers the same and no central guiding intelligence.

P2P represents the distribution of control from centralized IT departments out to individual users.

By putting control of network resources and addressing directly in the users' hands, P2P creates significant challenges to the traditionally centralized IT department, and to the current "intranet + firewall" networking model of enterprise security and control.

P2P is harnessing the capabilities of networked devices.

While P2P's initial impact has been to discover the untapped resources of distributed PCs, P2P will also change how we interact with all kinds of devices, many of which today are tethered to PCs. P2P provides network infrastructure for these devices to communicate more effectively and more directly with one another.

P2P and Web Services are closely related and may converge to become a single category in the future that signifies the effective use of distributed resources.

The direction of P2P and Web Services will be colored by corporate wrangling between Sun and Microsoft to establish open or de-facto standards for network infrastructure. Sun's strategy is one of complete openness and standards conformance. Microsoft's aim with .Net, HailStorm and Passport is to bring all of its various properties under one roof, with the customer placed squarely in the middle.

We consider the key application areas of P2P to be file sharing, instant messaging, distributed computation and P2P groupware. Most of the functionality found in P2P applications today will become more generalized as part of application frameworks or P2P infrastructure provided by Sun or Microsoft.

FILE SHARING

The future of consumer file-sharing services lies with licensed, subscription-based services. Fully decentralized services have a better chance of surviving legal attacks than do centrally managed services such as Napster.

P2P content delivery networks have a tough road ahead; they must deliver solutions that cost less and are comparable in service to "classic" content delivery networks such as Akamai.

The only realistic business model in the short-term is file sharing in the enterprise, currently dominated by NextPage's federation strategy.

DISTRIBUTED COMPUTING

Due to early adoption in the life sciences and financial services industries, distributed computation has the greatest short-term revenue potential in the P2P space.

A wide range of enterprises will find significant cost savings and process acceleration using P2P distributing computing, particularly when compared with the administration requirement of traditional distributed computing clusters.

Substantial opportunities exist for enterprises to distribute work to commodity hardware for at least the next decade.

INSTANT MESSAGING

Developments in both P2P and IM point to the emergence of PIE. In a PIE-enabled network, resources at the center migrate to the edge, anonymous users gain identity and transient connectivity yields to presence.

AOL's AIM's hold on the enterprise desktop is vulnerable; long-term success is dependent on opening to third-party vendors. AOL needs a platform strategy to compete against Microsoft.

Compared to AOL, Microsoft has orders of magnitude better penetration on the enterprise desktop. Its ability to provide presence services for business can ride on OS penetration, giving Microsoft an unmatched edge in corporate presence management.

Given the investment by France Telecom, Jabber may become a major contender if it can successfully tap into the SMS namespace.

Market share for IM will be turned on its head, with Microsoft coming to control as much as 80 percent of the market and AOL's AIM reduced to 10 percent, unless AOL gets a clue fairly soon.

P2P GROUPWARE

P2P groupware will be more successful than client-server groupware at penetrating the enterprise; it will not, however, be a universally adopted tool.

Groove dominates the P2P groupware sector but faces an uphill battle because key groupware functions will be integrated into mainstream groupware applications.

A number of groupware functions will also be offered as Web Services.

Only two or three large providers of P2P groupware will survive; many of the current entrants will be acquired by larger players or exit the space.

INFRASTRUCTURE

One-third of survey respondents identify themselves as P2P infrastructure providers. However, Microsoft and Sun, and possibly Groove, seem to be in the best position to supply infrastructure and establish the open or de-facto standards on which to build.

As P2P infrastructure matures, applications that make use of distributed resources will be easier to design and develop, and they will become reliable and widespread.

P2P Players

Most of the P2P companies categorize themselves as Infrastructure/Platform companies (34 percent); others are File Sharing (24), Collaboration (10) or Distributed Computation (9).

There are more than 150 P2P companies, and we've surveyed nearly 100 of them. Most P2P companies are small and privately held; typically, when larger public companies enter the P2P space, they have an overarching framework to offer, such as Microsoft's .Net, or Sun's JXTA.

Approximately $560 million has been invested in P2P companies, 94 percent of that in the past 18 months. In spite of extremely tight capital markets, $165 million has been invested in 2001.

Many of the well-funded P2P companies target improving communication inside large corporations.

Collaboration companies have received 37 percent of the funding; file sharing, 31 percent; and distributed computation, 18 percent.

Most of the companies we surveyed are using XML. Twenty percent are using SOAP and almost 10 percent of the companies are incorporating features of Microsoft's .Net or Sun Microsystems' JXTA.

Back to: 2001 P2P Networking Overview