Distributed Virtual Supercomputer

There are many individual projects and tasks which will require some computer resources beyond the reach of most individuals on the planet. This project is to establish a portal and a community of dedicated Wikiversity enthusiasts who can develop familiarity, skills and expertise suitable in the medium term for setting up distributed virtual supercomputing at Wikiversity. This will allow the participants' computers in screen saver mode to support each others projects.

Example projects might be high quality rendering or post processing for a large group project or an ambitious mathematical simulation regarding a current topic of big science.

Step 1
Establish teams titled wikiversity.org at already well established BOINC project.

Teams established

 * Rosetta@home
 * Einstein@home

Other good candidate projects

 * Seti@Home
 * Predictor@Home
 * Many others here - like Screensaver Project

Step 2
Participants begin joining screensaver computing teams in large numbers and building familarity with environment.
 * Task 0 Join the Wikiversity Mailing list and help brainstorm this or use the discussion page here.

Step 2.1
People volunteer for specific development tasks or responsibilities and begin discussing details and expanding proposal.

Step 2.2
When sufficient interest has arisen consider a mailing list or sourceforge project style forum to allow software developers to coordinate with seed hardware administrators, contributing philanthropists and client projects at Wikiversity.


 * Our developers may wish to congregate at sites such as this for specific discussions in depth of the BOINC technology specifically and simultanously at our own local pages specialized in Wikiversity's needs and capabilities.

Step 3a
http://boinc.berkeley.edu/vcsc.php

The above link may be different. Last time I checked we needed one off line computer for a security key to establish a computing project. Also a seed server connected 24/7. Initial development might be handled by a low end server and 24/7 DSL or cable service.


 * At the above link it is estimated that a $10,000 server can support thousands (single digit) of computing nodes (desktop screensavers) or approximately a couple of bucks per processing node which probably runs 8 to 16 hours/day.

It looks like the BOINC team is helping us by solving the general problem for us and all other universities. It will still be a lot of work to setup and administer. We need to get started soon to help attract highly qualified and motivated participants to our science and technology programs.

Step 3b
Establish first local BOINC project. A good candidate might be an outreach program of some kind aimed at high schools. U.S. has a lot of high schools with a lot of computers sitting idle two thirds of the 24 hours day.

Step 4
Begin evaluating Wikiversity infrastructure and setting up teams interested in various aspects, layers or chunks of the problem of a distributed reliable grid to support a virtual university. Perhaps answer is for WMF to buy more servers. Parts of the answer may eventually involve large persistent online distributed datastores such as might be available with OceanStore or gnunet.

Step 5
Propose improvement tasks back to WMF for other Wikimedia projects.

Step 6
Setup specific operational virtual computer laboratory projects aimed at educational original research. Berkeley's Open Infrastructure Network Computing (BOINC) recommends experts allow at least several man months to get an initial development system partially up and running. Obviously it will take a merry band of volunteers partime quite a bit more effort and calendar time is subject to who and how many volunteer to learn and do what.

Open source software helpful to the project

 * Sandbox Server - place for coordination of a "Virtual Supercomputer" project
 * gLite
 * GPU, a Global Processing Unit - p2p internet supercomputing framework
 * NetWorkSpaces for Python - Assisting python in distribution of processes.
 * Hadoop
 * BOINC
 * CPU Share
 * paexec
 * GitTorrent