| Monday, November 5 |
So many ideas and so many technologies are swirling around P2P --
decentralization, distributed computing, web services, JXTA, UDDI,
SOAP -- that its sometimes hard to see order in the chaos. If we look
past the labels and the individual applications, though, we can see
one thing clearly: The twin revolutions of the PC and the internet
collided in 1994, and for half a dozen years after, browser+server was
the indispensible internet architecture.
Today, we are witnessing a Great Re-wiring, a chance to reconsider how
the world's devices connect to the internet and to one another outside
the browser+server framework, and no matter what label we choose for
it, this re-wiring is transforming the network and how we use it.
To Be Announced Soon.
To date, all the roundups and analyst reports on P2P have had a stilted,
shoehorned feel. That's because the wrong metrics are used to categorize and evaluate P2P tech. On the one hand are the financial people, who distinguish among P2P offerings based on their
markets and possibilities -- B2B, B2C, subscription, licenses, etc. On
the other hand are technical people who divide up the world according to what
the software is used *for*: file-sharing, instant messaging, distributed
supercomputing, etc etc.
Doctorow believes both of these approaches are flawed, and miss the point
entirely. The way to group and understand P2P offerings is to examine what
makes them *cool*. In a Niftiness Hierarchy (NH), P2P companies might
shake out like this:
Letting the contents of the desktop be shared to the world, extending DNS to
the desktop
Puncturing firewalls and engaging in guerilla networking
Storing and transferring files without restraint
Collaboration across the org chart, across the world
Cheapo renderfarms, big science gone dingo, homebrew biotech
Modeling industrial decision-making on TCP/IP: I know how to complete part
of this task, I know what the task looks like when it's done, and I know who
the next person in the chain is, but I don't know the whole thing.
Solving Metcalfe's Paradox by allowing users to provision the resources they
consume.
Relay networks whose connections reflect trust: as with OC Folders, where
self-organized topologies take the place of directories
On this panel, their proponents explicate three metaphors for P2P, and an attempt is made to find the grand, unifying metaphor of P2P. In other words, we're gonna find the koan that awakens P2P enlightenment in even the most clueless of seekers.
The three metaphors:
Each panelist spends 1-5min introducing their metaphor. The next twenty
minutes are devoted to talking about the strengths and weaknesses of each metaphor. Remainder of time: Audience participation (suggesting new metaphors).
90% of the Internet is invisible. P2P is significant as the Internet for the
Rest of Us.
The stand-alone PC is an amoeba. The P2P Internet is a cluster-organism.
Successful cluster-organisms exchange messages (XML); sucicidal ones
exchange DNA (mobile code, Java). The P2P strategies that succeed will be
based on messaging, not RPC.
The thing that's exciting about P2P is Metcalfe Squared. Not only does the
network gain utility with each additional user, it also gains capacity. IOW,
each user provisions the resource he consumes. Forget the Cornucopia of the
Commons -- this is a Commons where the sheep shit grass.
Peer-to-peer (P2P) computing is an old idea made new through the ubiquity
of powerful computing platforms and the Internet. The original P2P computing evolved from the ARPANET where "peer hosts" -- orginally DEC PDP
minicomputers -- became the genesis for distributed defense networks such
as the World-wide Military Command and Control System (WWMCCS).
The US military is rapidly attempting to adapt P2P under the concept of
network-centric warfare where the computer is a part of everything, and
everything is part of the net. "Everything" includes individual soldiers,
their vehicles and aircraft.
This reflects the wider trend in the commercial world. By 2005, the average
COTS PC will come with a 4 Ghz processor, 500 GB of storage, 256 MB of RAM,
Gigabit ethernet, and a broadband connection to the web. PDA's will only be
a few steps behind with wireless connectivty over 3G networks.
In essence, we will live in a world with a vast number of networked
super-computers running a wide array of applications with national security
implications -- from codebreaking to large-scale simulations.
P2P will have major implications on how US forces will operate in conflict
and on the speed of decision in foreign policy. Moreover, P2P will create
new possibilities and vulnerabilites that we are only begining to
understand.
The Evolution of Gnutella
Part I: The Past
Gnutella has made great strides in the past 8 months. The basics of file sharing have matured nicely, and network usage has grown rapidly. Much more significant are the several pieces of the Gnutella net which are just being released now. Structured queries, software which allows web sites to respond to Gnutella queries, and browser integration all combine to give Gnutella a completely new set of functionality that may allow it to live up to its initial hype. LimeWire going open source provides a common base for development that should allow others to innovate without having to reinvent the wheel. The Gnutella net is now open to a much richer variety of possible information which can be passed and shared. Commercial enterprises can now use the Gnutella net to field inquiries broadcast by consumers. Although many chicken and eggs problems need to be surmounted, the next year should see the first generation of businesses using the Gnutella net as a valuable resource to generate profit.
The Present
The future
How you can get involved
What makes Gnutella good
Since law is often formed as a way of approximating ideals, Gonze discusses what the relevant ideals are. The attempts to impose material scarcity on the digital world compare to the unibomber's attempt to recreate an agrarian society. Gonze explores whether Richard Stallman said anything relevant in the Gnu manifesto, whether open source and free software are different moral stances, and where the prevailing sentiment among P2P developers - "Don't fight it" - fits in.
The Buddha-like calm of Napster fans regarding the potential for
Napster to cause suffering is a strange thing, as is the indifference of the RIAA to the nuttiness of creating scarcity for its own sake. If this were food being copied instead of music, would there even be an argument?
Well, yes, there would; we all manipulate scarcity to make a living. We seem to be in a Star Trek (Kirk Generation) moment, where we have to decide whether to pursue utopia or business as usual.
Mark Lucovsky, a distinguished engineer at Microsoft and the architect of the .NET My Services, discusses several Microsoft initiatives for building distributed services. Lucovsky outlines the .NET My Services programming model and how it is used for both peer-to-peer and distributed applications, as well as some related developments in peer-to-peer infrastructure.
| Tuesday, November 6 |
Given that the web is all about massive connectedness, what are the driving principles behind the evolution of the web and what do they tell us about the harmonisation of web services, peer-to-peer computing and wireless connectivity? Simon Phipps explores the principles behind the evolution of the web, position initiatives such as UDDI, SunONE and JXTA in the evolutionary scale, and peers into the future of services on the internet.
This session is an introduction to the JXTA platform: design concepts, protocols overview, features, applications, and future direction. JXTA technology addresses the need for an open, generalized protocol that interoperates with any peer on the network, including PCs, servers, and other connected devices.
This session is for the advanced P2P developer. It takes an in-depth look at the JXTA platform from Sun Microsystems, noting the scope and limitations of the platform. Schneider discusses how to fill some of the gaps, and identifies many of the challenges that face JXTA developers. Examples are given using the Java bindings. Since the JXTA platform is still in its infancy, the details of the session will emerge as the platform grows.
Sun and several key partners discuss opportunities for developing services and applications using JXTA.
Internet Service Providers will face increased pressure to control the information that passes through their pipes to the Net, if peer-to-peer computing disperses the usual suspects. Legal or market threats to ISPs, and from them to users, could reimpose the centralization P2P was supposed to eliminate. This panel considers how users can work with these unnoticed intermediaries to preserve unmediated Internet communications.
People often assume that "peer-to-peer" or "distributed" always means
decentralized. But it doesn't. While successful systems like Gnutella
and Freenet take decentralization as a primary design goal, others
like distributed computing companies are largely centralized. And most
current peer-to-peer architectures are somewhere in-between: whether a
hybrid approach like Napster, a hierarchical system like DNS and NTP,
or a complex system such as Groove.
Nelson talks about centralization and decentralization as a basic
design principle, relying heavily on examples from current systems. He
illustrates the four fundamental patterns of building distributed
systems (pure client/server, ring, hierarchy, and pure
decentralization). These approaches have different benefits and
drawbacks, including scalability, manageability, complexity, security,
and resistance to legal or political intervention.
Nelson concludes
with an in-depth exploration of two systems he designed
with similar goals. One system, Hive, was fully decentralized.
The other, Popular Power, was fully centralized. Each approach has
advantages, disadvantages, and lessons to learn. In the end, while
decentralization offers exciting flexibility and scalability,
centralized systems are often easier to build, manage, and understand.
REST (REpresentational State Transfer) distills the Web's architecture into a general framework. The Web succeeded by defining a small but expressive set of methods -- GET, POST, etc -- that can act on an unlimited set of resources, represented as URLs. The resulting system is stateless, loosely coupled, and scales better than RPC implementations such as DCOM, CORBA, and RMI. What does REST mean for Web Services? Does REST describe a way for building intra- and inter-application communication that is as flexible as the Web itself, while avoiding the brittleness and coordination costs of RPC?
If we can find ways for content creators to make a living regardless of unauthorized distribution, is it OK? Or is it piracy no matter what the outcome? In the hope of moving the conversation about filesharing beyond the standard rants, this panel discussion attempts a thoughtful dissection of a messy issue.
What separates stand-alone P2P applications from the players who will leverage P2P into web services/peer services? The O'Reilly Report on P2P suggests the answer is PIE -- Presence, Identity, Edge. Join the report authors as they define and discuss PIE and look at what it takes to become a PIE player.
What the web did for business-to-consumer interactions, web services are poised to do for business-to-business interactions. Built on top of new and emerging technologies such as HTTP, XML, SOAP, and UDDI, web services are the next horizon for e-business. The key to reaching this horizon is a common program-to-program communication model that changes the rules of integration both within and across new and existing enterprise systems.
This presentation will cover web services: what they are, why and how they are changing program-to-program communication, and how they are transforming businesses. It will explain the key web services standards, their adoption in the industry, and some of the key emerging directions. Finally it will give a brief summary of IBM's current and future product support for developing and deploying web services.
| Wednesday, November 7 |
Congressman Rick Boucher (D - VA), leading architect of federal policy for the Internet, discusses the impact of web services and peer services on the Internet legislative scene.
Larry Lessig describes the changes in law and technology that threaten the innovation commons created by the Internet.