|
3-19-2008 This page is undergoing editing
Imagine your brain as a canister filled with ink.
Yeah, now think of your body as the pen where the ink resides.
Fuse the two: KAPOW! What are you now?
You're the human magic marker. Won't you please surprise my eyes.?
It's in your nature, you can paint whatever picture you like.
No matter what Ted Koppel says on Channel 4 tonight.
So modify this third rock from the sun.
Excerpted from "Redefine" by Incubus
Introduction -- What is Hangar18 ?
Hangar 18 is an attempt to develop Open Souce technology and solutions in
the area of Distributed Computing. Our core technology base is built around
Plan 9, Inferno, and Linux. However, we do not limit particpants to
just these operating systems. We do not believe that any single OS will be
the solution to all computing problems. There are two basic aspects to the
way we organize our efforts and shared resources; small world networks
and a tit-for-tat access policy.
Small World Networks
A small world network is a mathematical model of the way humans build their
social networks. The basic idea is that while each person knows only a
relatively small number of people, when we look at the people these people
know we discover that the distance between any two persons isn't that
great. This is aspect is known under a lot of names, "The Kevin Bacon Game"
is a well known example. As research into this area has grown over the last
few decades a much wider application for these models has been found. While
small world networks are initially hard to setup they tend to be very robust
and long lasting.
The particular approach that we pursue is to limit the number of connections
between any two nodes in the Hangar 18 network within two broad goals.
Number of Connections Per Node
We ask that any node have no more connections than the ln() of
the total number of nodes.
Common Shared Connections
We ask that no two nodes with a direct connection should share
any other nodes in common. This prevents a lot of redundent
connections. These can have two effects, they create a lot of
parallel paths that complicate routing and they make the network
as a whole more fragile to single points of failure.
Tit-for-Tat Access Policy
One of the biggest problems of shared resources is the Tragedy of the
Commons. This is the situation where a few people hog all the resouces
without providing a lot to the community in return. To avoid this we turn to
Game Theory and a widely studied problem called The Prisoners
Dilema. Without going into the details of what this means exactly we can
summarize; if you want to access Hangar 18 resources then you must also
provide Hangar 18 resources.
Where Will Distributed Computing Be Important?
Distributed computing wil have a great effect on the way we all use our
computing resources. In particular issues of the user environment and how
distributed computing can be made ubiquitous. This includes multi-player
games, distributed real-time computing, shared media libraries, multi-media
conferencing, mobile global sign-on, wireless and packet technology,
cryptography and anonymity, digital cash, robotics, distributed multi-agent
systems, etc.
The Power of Open Source
The Open Source movement is a little over 20 years old. It has done a
great job at moving existing technology into areas that commercial
applications would have a hard, if not impossible task at doing. It has also
helped limit the impact of monopolizing commercial interest. This has had
two effects. The first is keeping the availability of technology to the
individual alive and well. This is important because it keeps innovation
from being locked into a few small cadres of commercial effort. The second
is to help control the market entry costs so that the resources needed for
new commercial efforts is not too high. Both of these are important to
keeping a free market society healthy. The impact of the GNU tools and Linux
are primary examples. As a counter-example of how commercial approaches
stifle innovation is the history of Plan 9 and its adoption by the computing
community as a whole.
But this is only the beginning, one observation, some would say complaint, of
Open Source is that it has not been as innovative as hoped. What does it mean
to be 'innovative'? We believe this means to solve old problems in
fundamentaly new ways and to discover and resolve new issues and tasks that
weren't possible or had prohibitive costs under the traditional technology
license approach.
Another aspect of Open Source is it's ability to keep technology available
to all instead of bottling it up so that only a select few have the
opportunity to understand and manipulate the fundamentals of that
technology. This is one of the problems if conventional intellectual
property approaches, it is inherently monopolizing. Unfortunately the vast
majority of Open Source developers and users are not looking at new
problems, they simply want to solve their existing problems for free.
This is not the primary goals of Open Source as envisioned by Hangar 18. We
hope to explore and develop fundamentally new approaches and at the same time
create a intellectual property framework around them that will prevent
monopolization.
What is it we're trying to do?
The primary goal of Hangar 18 is to develop a truly distributed computing
infrastructure. This means global sign-on, distributed processing, a variety
of resource name-spaces, layers of communications infra-structures that
support various ranges of point-to-point communications, and hopefully much
more.
The current goals are to create a multi-tiered public resource that has the
following features:
Global Sign-on
meaning that you take your laptop anywhere on the planet and if you
can get a route back to your core authorization servers you have
access to all your resources as if you were sitting at home.
Contingent upon the base bandwidth of your connection of course.
Process Clouds
Access to a processing cloud that allows seamless expansion of
processing power as needs grow. This means that your computer is no
longer that single box in front of you. That machine becomes a
input/output processor that handles just the real-time interface.
All processing takes place in a distributed cloud of machines which
can be located anywhere on the planet. If one machine goes down the
distributed namespace stores sufficient context or state to
switch the process to other resources without intervention or even
the user or process sponsor being aware that a switch has happened.
Process Namespaces
A namespace, similar to a directory tree, that organizes resources
around the planet. This means that you could have access to many
terra bytes of storage as a regular resource. By mapping hardware
as well as traditional file and program resources in this manner
makes issues like multi-cast trivial. For example, you and your
friends could mount your sound cards under your /dev directory
and then cat your /dev/mic output to /dev/sb* and every participant
would recieve the data flow automatically.
Security
Security through the application of encryption technology at all
levels, from the network infrastructure to file storage in the
distributed namespace. One approach would be to break up all files
into small pieces. Store those pieces in many locations in the
global namespace blinded through an anonymizing layer so that those
providing storage would not know what was actually being stored or
even where. With a sufficiently large group of servers using
parallel storage the concept of 'where a file is stored' becomes
rather fuzzy. The social implications of non-censorable content are
an open issue at this time.
Distributed Services
A lot of current services and features on the Internet and within
conventional operating systems rely on centralized authority or
access. This is not a requirement but a convenience within the old
model of resolving issues. We hope to create an environment with
distributed services ranging from name resolution, which allows
free form domain name creation, to anonymous free running
search engines. Another extension of this goal is the install
once, run everywhere model. The impact of such approaches to
the bottem line of business is an uncharted territory.
Groupware Applications
The impact of working in groups under distruted resource model
will be immense. Music performance and gaming are obvious examples.
Wireless Networking
The utility of wireless technology looks much more reasonable
when looked at under this model, as compared to the traditional
commercial approach. This means that a variety of wireless
technologies can be used to create various 'layers' of connectivity
in a community. One uses 802.11 a/b for short range line of sight
connection; for example within a couple of blocks of your home.
Then switch to 802.16 and 900MHz unlicensed radio for more mid-range
connectivity; perhaps from 5 to 30 miles of your home. By judicous
use of 'wireless bridging' technology this means that one could
roam the city and stay in touch with your home network at all
times. Thus one could still have connectivity to traditional
resources through your home or personal services.
Emergency Services
Emergency services would benefit from such a community
infrastructure in time of crisis as the community would already
have a social framework of sharing in place and the technology
coupled with deep experience in how to use it.
Research and Education
The applications of distributed computing and wireless technology
are another area that is only beginning to be understood. The
impact of distributed networked computing to fieldwork in a
variety of industries will be revolutionary with regard to the
cost of doing business and managing the impact on the community
and the environment.
How To Participate ?
Participation is on a tit-for-tat basis. What this means is that in order for
you to participate, you find someone already participating, and arrange an
agreement to share resources. In other words, you gain access to the other
participant's resources by providing resources of your own.
The primary contact mechanism is the Hangare 18
mailing list. This is how you find people to share resources with. It's
also where notices of meetings and new resource availability are made.
We hope that you can make use of these resources and technology to meet your
own goals. Your participation will help others as well. The goal is that
this sharing of experience will change the world for the better.
The simplest way to participate is to start a local Hangar18 chapter, perhaps
even putting up resources for your local community. We have weekly social
meetings in Austin (usually at a restaurant), and other flexibly timed meetings
for working on specific projects. While user groups often have large
presentation format meetings, we also prefer small gatherings focused on
hands-on projects, where everyone learns by working together.
Hangar 18 Participation How To
Projects
- Current Mailing Lists
- Plan 9 Snapshot (June 8, 2003)
This document provided a skeleton for a talk to the
Robot Group in Austin.
Austin Robot Group Plan 9 Talk - Outline
(.doc)
A step-by-step guide on how to setup a basic Plan 9 I/O-Authorization
server. The core to full participation in H18.
A Gentle Introduction To Plan 9 - How to
build a basic Hangar 18
This is a short Plan 9 rc script that is needed for our intro.
h18_lnksrc.rc
- A Distributed
Network of Computing and Communication Resources, shared on a
tit-for-tat basis. The "tit for tat" part means that you get access
in exchange for providing something; like getting a multi-course
dinner party in exchange for bringing one dish to a pot-luck. This
project that will provide the backbone for sharing information and
computer resources on other projects We are using plan9 because
it seems well suited to this and has exciting capabilities; we will
not make a particular OS necessary to participate.
- Open Air
Optical Network, consisting of a laser emitter and a detector
attached to the serial port of computers. The main goals were (in
the near term) to refresh and teach ourselves about electronics,
and (later) to provide more linkage into the Hangar18 network.
- Community
(Guerilla) Networking, consisting of a link between two or more
Hangar18 nodes using 802.11a/b, 802.16, and 900MHz.
- Distributed
Gaming, the use of Squeak
(Smalltalk) (or other systems) to allow users to create their
own realtime, distributed gaming environments on Linux, Plan 9, and
other operating systems.
- Distributed
Electronic Cash, an exploration of the applications of computer
technology and economic (monetary) systems. This includes Chaumian
Cash, Distributed Bearer Bonds, and similar efforts. The
Cypherpunks are one
non-professional group exploring these applications.
- Distributed
Databases of various types and uses. Automatically map wireless
spectrum and update an online map everywhere you go; link many small
CAD-equipped telescopes to look for asteroids; or anything else where
there is benefit to the aggregation of numerous small amounts of
data.
- An Open Technology
Laptop, taking advantage of recent advances such as low-power
mini-itx motherboards to
allow individuals to build laptop or PDA-like devices of their
own design.
-
Distributed Autonomous Robotic Wars (DARWars): Intended to be a
means to explore distributed computing applications in swarm
computing, autonomous agents, wireless communication, and remote
real-time control (waldo's and proxies).
Copyright 1983 - 2008 All rights reserved.
Permission to use components of this website for non-commercial use is
granted.
|
