Introduction
There are numerous initiatives
underway to create the Next Generation Internet. Sometimes this
is referred to as the Grid. There are also a number of company
specific labels mentioned in the public domain such as Microsoft's
.Net, Sun's SunOne and IBM's WebSphere. Microsoft's vision is
one of a connected web where things happen automatically, rather
than users having to make their connections and pass information
onto other people in a manual way. The World Wide Web Consortium
(W3C) also regards the current version of the Internet as a prototype,
and is working to produce something that it calls the Semantic
Web. Underlying these developments is Extensible Markup Language
(XML). XML already brings structure to the net and is particularly
useful for user interfaces and for making assertions and assumptions
about the meaning and relevance of data.
Agent Technologies
The Internet currently provides
the wiring. What is needed to provide intelligence in the Internet
is software technologies embedded in the Internet itself. One
form of software are agents and multi-agents that are capable
of acting autonomously for their owners (individuals, companies,
and others) - pursuing goals, modifying their behaviour, co-operating,
negotiating, etc. These agents would interact with other agents,
dynamically. Numerous efforts are being directed at agent and
multi agent systems, world-wide. Open issues and challenges are
how to describe and reason about services; handling different
languages; enabling brokerage between different agents; and scalability.
The World Wide Web Consortium
(W3C) has created guidelines for people creating agents. W3C
also has responsibility for XML, which will eventually allow
agents to use semantic information to perform and steer their
tasks
Peer-to-Peer
Networking
The agent based approach attempts
to put intelligence into the Internet. An alternative one is
to use the intelligence that exists at the periphery of the Internet,
that is in the millions of personal computers that are connected
in offices and homes to the Internet via Company servers and
Internet Service Providers. This field of activity is known as
peer-to-peer (P2P) networking. The technology for this type of
activity has largely emerged from the USA over the past four
years. Names associated with this field are Napster, Gnutella,
Flycode, Freenet, GoneSilent and so on. There are also vendors
(e.g. Groove Networks) selling development tools for business
applications of P2P networking.
Napster, the music download
web site is an example of P2P technology. Napster provids a directory
of music files located on many personal computers throughout
the world, plus a piece of software that allows people to download
files. This system works as follows. Anyone with music files
on their computer can publish on the Napster server, a list of
files they are willing to share. The server matches up requests
for files with the list of providers, but the files themselves
are transferred directly from personal computer to personal computer.
These files do not pass via the Napster server as they would
in the traditional Internet model that we know today. This is
possible because Napster solved the technical problem of actually
how to work around the structure of the Internet and allow computers
that do not have domain names to locate each other and exchange
specific files.
Although still in its infancy
from an industry acceptance point of view, (in the case of Napster
for obvious reasons), this type of technology is already dated.
A system called Gnutella leads to the same end result: obtaining
music free of charge, but without the central server directory.
Gnutella users just download the software needed to operate the
system. When someone makes a request for a file, the request
is examined by other personal computers and relayed onwards to
yet other personal computers down the chain, until the requested
file has been located.
FreeNet differs from Gnutella
in two important respects. First, the information is encrypted,
so the person who originally put it on to FreeNet cannot be identified.
Once the information has been posted, it moves randomly to another
computer. The user of that computer will not know what information
is on the machine. Because FreeNet has no central directory,
a search engine looks through the entire network each time anyone
seeks a file. The second difference is that FreeNet tries to
be efficient. Unlike Gnutella, it looks out for popular files
and ensures that a number of copies exist in various places.
FreeNet will also move the information close to the place where
it is in demand. This helps ensure that computers containing
the information are not overloaded with requests. This also enables
access to files even if the originating computer is off-line.
There is also growing commercial
interest in these technologies (this is represented by the involvement
of "Blue Chip" companies such as Siemens, Intel, IBM,
and others) and tool and technology vendors supporting reputable
applications are also beginning to emerge. Most of the focus
in the area of P2P technologies at the moment is on sharing copyrighted
digital content (both legally and illegally). There is also a
growing interest in building faster search engines using these
technologies and expanding the range of sources searched by adding
computers on the fringes of the Internet. Industry interest also
lies in more effective sharing of information and computing resources
across the enterprise.
There are already two well-publicised
examples of P2P networking technologies being used to enable
scientific studies that would have been previously impossible
owing to time and money constraints. Both of these involve the
use of home personal computers, networked using P2P technologies,
to carry out scientific analysis. The first of these was the
Seti@Home initiative which involved using the power of more than
two millions PC to analyse radio-telescope signals for evidence
of extraterrestrial intelligence. The second, a project situated
in the health area, begun in March 2001 and involves using PCs
to screen 250 million chemicals for their anti-cancer activity.
Many developers in the P2P
field have adopted the open source philosophy. Also the field
does have reputable interests and potential applications, and
is not just used for copyright infringement activities such as
free music file sharing. The technology does in fact appear to
be ripe for application, providing a powerful means of both expanding
the information resources available on the web as well as searching
for those resources.
Conclusions
In connection with the software
technologies such as agents and peer-to-peer networking, a major
issue is that of security, trust and privacy. This is already
a big problem and these new technologies seem to be just creating
even bigger security, trust and privacy problems. With regard
to agents and services there is a need to answer questions like
how an agent can describe the information and resources at its
disposal and how an agent specifies the tasks it has to carry
out. With regard to agents and brokerage, the question is how
does an agent find another agent capable of doing a task, how
an agent can resist other agents ganging up on it, and how to
find a best fit between goals and identified services. Scalability
is also a difficult area because of problems like feedback, resulting
in instability or undesirable behaviour.
The coming years will certainly
see significant developments in Internet technologies. What we
have today is only a rudimentary system. The Next Generation
Internet will be more sophisticated and offer greater potential
to do entirely new things. Next Generation Internet will however,
also pose even more severe problems for security, trust and privacy. |