Peer-To-Peer Computing and Your Business
Imagine a computing utopia. Imagine a world
in which bandwidth is not a constraint. Imagine that peer-to-peer
(P2P) software has been perfected and is plug and play. Imagine
that every computer in your organization is able to contribute
its idle resources to whatever purpose you choose, and if you
need more resources, they’re easily, cheaply, and instantly
available. What would you do? What kinds of business problems
could you solve with virtually limitless computing resources?
Wall Street’s answer is an application that
crunches fantastic amounts of data to support the trading of
securities. DataSynapse, which develops P2P software that links
computers together to solve financial problems, completed a
test calculation for J.P. Morgan that normally took eight hours
overnight in a matter of minutes. According to the company,
12 high-performance financial applications are being optimized
over two farms of 400 workstations and 250 PCs.
Computing power isn’t the only target for Wall
Streeters. Like many of us, they also need help managing the
enormous volume of information available, and distilling it
into actionable insights. P2P startup WorldStreet has released
WorldStreet Net, a P2P information management and publishing
application targeted at the financial industry. Developed with
WorldStreet’s eXtensible Integration Architecture (XIA), an
XML-based integration methodology, WorldStreet Net enables users
to capture, manage, and modify their business relationships
on-line. The application also offers structured message types
that use standard financial industry terminology. Examples include:
trade ideas, investment packages, internal notes, analytics,
events, research, news and notes, recommendations and earnings,
reports, and dialogs. Every user can be a publisher and can
create new workflows using the system.
These examples are just the beginning. Hundreds
of P2P companies have sprung up to take advantage of the estimated
10 billion MHz of processing power and 10,000 terabytes of storage
available in the world's Internet-connected PCs. P2P is a major
trend, and it’s accelerating at Internet time. Although it was
an obscure concept even a year ago, P2P has made the front pages
since the release of Napster, a music file sharing service,
in early 1999. Now, Forrester Research’s CEO George Colony says:
"This technology will come along and kill off the Web.
. . . And that judgment day will arrive very soon -- in the
next two to three years, not 25 years from now."
P2P As ASP Descendant
P2P networks can be seen as the ultimate flowering
of the Application Service Provider (ASP) concept. The core
idea behind ASP is that applications can be made available as
a service rather than an installation. Businesses can contract
for application capabilities from companies whose core competencies
are to produce and maintain the service. This yields efficiency
since such specialization can make the ASP the low cost provider,
and this not only can save a business money, but relieves it
of the burden and cost of maintaining technological currency.
B2B P2P networks will take these concepts further
by removing the need for the application to live in a particular
location, on specific hardware. A P2P application can obtain
resources from anywhere on the network. An application can get
disk space from here, computing cycles from there, specialized
application services from over there, and network bandwidth
from yet another place. It will be possible to configure an
entirely new capability in real-time through the assembly of
parts available on the network.
P2P resources currently are treated as a low
cost commodity, representing, as they do, idle capacity. But
in the future, these resources’ value will be recognized, with
an accompanying re-evaluation of their cost. ProcessTree, a
P2P CPU-cycles vendor, may soon offer a standard CPU-year, currently
defined as a 400-megahertz Pentium II operating full-time for
365 days, for about $1,500, less than a fifth the cost of equivalent
time on a supercomputer. In fact, you can beat the performance
of a $50-million, massively parallel research machine by convincing
1 percent of the people on the Internet to donate their unused
computing cycles. Unbelievably, the resources currently available
are dwarfed by their possible applications. According to United
Devices CEO Ed Hubbard, genome applications alone could soak
up all the Net's spare computing power for the next 50 years.
P2P Today
Currently there are four categories of P2P
providers:
Disk sharing – Companies like Napster,
Gnutella, and Aimster organize networks of distributed disk
storage. They provide search or indexing capabilities and file
upload and download capabilities. Business use of disk sharing
networks will likely be limited to behind-the-firewall or other
types of dedicated uses due to concerns about security. Because
of redundant storage, even the consumer networks are highly
reliable, however.
Processor sharing – Companies like Popular
Power, ProcessTree, and Applied MetaComputing and non-profits
like SETI@home distribute pieces of data and applications to
net-connected PCs, who work on the application when they are
idle. A more sophisticated version of this technique involves
using excess processor power while the machine is in use.
Bandwidth sharing – Companies like 2AM,
Kalepa Networks, and Nextpage use P2P to help distribute high-bandwidth
multimedia content via caching. Rather than streaming the video
for a movie trailer from a single high-bandwidth site, , for
example, these companies can distribute the trailer to participating
peers and direct users to those sites closest to them.
Content sharing and publishing – Companies
like WorldStreet, Consilient and Flycode offer peers the ability
to find relevant content and publish their own. A related niche
is distributed search engines such as Pointera which combine the power of peers to search the Web.
The P2P Future
Business P2P networks will need go beyond the
current artificial division of labor between file sharing and
idle resource maximization to provide a rich, collaborative,
integrated marketplace of business capabilities both within
the enterprise and externally. As they evolve, these networks
will develop the following characteristics:
Dynamic – Communications are interactive,
extensible, flexible, and easily reconfigured in real-time.
Systems link people, applications, computer systems and devices
such as cell phones, cameras, and printers.
Real-time – In many cases, sub-second
response time is required to deliver rich, timely, personalized,
on-demand information. Store and forward methodologies may have
a place in the P2P network, but most communication is immediate.
Collaborative – Both people and applications
need to work together to deliver value. These collaborations
must be secure, must support any number of participants, and
must enable the discovery of new networked resources.
Structured – Network services support
the vocabulary used by the business. This may extend to support
for local languages as well. Applications support the business
activities in a manner appropriate to the industry. In addition,
searching for information, services, or people is supported
in a structured manner through standard metadata.
Relevant – Information and services
are timely and focused on the participants' current business
needs. Users can personalize filters to customize information
delivery.
Service-based – Network capabilities
can be obtained and configured at a moment’s notice. New business
applications can be assembled on the fly by integrating new
capabilities into existing workflows, systems, devices and applications.
Cost effective – The network reduces
the costs solving business problems as well as of establishing
and maintaining on-line business relationships. Services are
provided by low cost specialists and are easily integrated into
the core business of a company.
Client focused – Services and capabilities
all can be easily personalized or otherwise adapted to the business
purpose.
It’s Not All Roses, Yet
There are still a few significant challenges
on the road to the ultimate P2P network. Chief among them is
security. Many, if not most, of the P2P applications currently
available do not implement any security. It’s hard to believe
that businesses will trust even tiny pieces of their information
and processes to anonymous PCs and their users. Encryption and
other security measures may not be enough to change this feeling,
at least in the short term.
Another facet of the security problem involves
administration. Traditional access control methods usually depend
on a central authority to administer policy. How would this
be implemented in a P2P network? Would all peers need to communicate
with a central authority for authorizations?
Of course, unlike in our imagined utopia, today
bandwidth is a consideration. To bring about the ultimate P2P
network, businesses’ networking capacity will need to be improved
exponentially. While 100Mb switches are now becoming more common
within businesses, Internet connections average T1 speed or
below. The advent of new networking schemes over the next few
years, including fixed wireless and the next generation Internet,
will go a long way toward removing this obstacle.
For public P2P networks, there will be a problem
of what to do with those who act to decrease the value of the
network, such as so-called leachers and spammers. Leachers pull
more value from the network then they provide. For example,
a user who downloads many songs but does not share any files
for other individuals to download is a leacher. Currently, a
large percentage of Gnutella network users are leachers.
Spammers propagate unnecessary and unwanted
messages, consuming attention and resources while providing
no value. We’re all probably familiar with email spammers. In
current P2P music networks, some record companies are uploading
apparently legitimate music files that actually contain only
white noise. Music companies are doing this in order to lower
the value that users can reliably extract from the network and
thus lowering the appeal of the network.
Since a P2P network has no central authority,
it could be difficult to minimize the impact of abusers.
Another huge problem for P2P networks involves
metadata, or data about data. In order for huge, somewhat chaotic,
P2P networks to flourish, there needs to be some standardization
in the way content and resources are described and identified.
Current brute force Web search engines give an example of what
could happen at a larger scale to P2P. How many times have you
typed in a word, Cobra for example, and got back irrelevant
results (information on snakes, a car, a malt liquor)? Probably most of the
time. If the proper metadata was available to allow search engines
to properly categorize content, you could specify the context
of your search and receive relevant results. The value of P2P
content networks will be greatly diminished unless a metadata
classification standard if developed.
What You Should Do
If your business has a need for supercomputer-class
capabilities, consider investigating one of the distributed
computing P2P vendors. See our directory
of P2P vendors for some places to get started.
If your business has a need to organize information
and enable knowledge publishing, consider looking into one of
the content and collaboration P2P vendors.
If you have massive multimedia content publishing
challenges, consider using one of the P2P content distribution
networks to alleviate the load on your servers.
If you have disk space challenges, consider
enlisting one of the P2P disk sharing vendors. And if you are
a Storage Area Network provider, consider worrying a lot about
P2P.
If your business deals with consumers in the
music or other entertainment field, you absolutely need to understand
the impact of the Napster revolution on your business. Below
are some resources to help you get up to speed on P2P computing.
And if your network person comes to you with
a requisition for new networking capabilities, sign it. Those
with antiquated connectivity will be left behind in the P2P
revolution.
If you decide to take the P2P plunge, you’ll
need an accomplished development partner like Geneer to help
you navigate this rapidly changing technology market. Geneer’s
vast experience in the ASP market can help you leverage the
P2P revolution.
Resources:
For a list of P2P vendors, see our directory.
The following resources can help you develop a good grasp of
P2P.
P2P Terminology Glossary
www.exocortex.org/p2p/terminology.html
Scientific American
www.sciam.com/2000/1100issue/1100cyber.html
TechRepublic
www.techrepublic.com/article.jhtml?id=r00520001227gcn01.htm
Peer-to-Peer Working Group
www.peer-to-peerwg.org/
New Productivity Initiative
www.newproductivity.org/
Peertal
www.peertal.com/
O'Reilly Publishing P2P Devcenter
www.oreillynet.com/pub/q/p2p_alpha
Global Grid Forum
www.gridforum.org/
. 