RS: structured mapping of real p2p internet infrastructure<http://groups.google.com/group/openmanufacturing/t/50f49ae67c75b034> Giovanni Lostumbo <[email protected]> Jan 09 08:39AM -0600 ^<?ui=2&view=bsp&ver=ohhl4rw8mbn4#12d6fbc88b29d451_digest_top> Hello, As there are many ways do map an infrastructure, one idea I've thought about was having all p2p software and operating system package either pre-installed on a computer and/or available on a GNU GPL distro (like or using aspects of GNUnet). The attached illustration is based on a partial-mesh topology 3 links/node in a 9 computer network (I merely included duplicate software lists at each PC to illustrate the example of modular and compatible PCs in any network). This could be scaled up, but might need a different software that can readily adapt to different topologies. Giovanni Lostumbo Giovanni Lostumbo <[email protected]> Jan 09 03:36PM -0600 ^<?ui=2&view=bsp&ver=ohhl4rw8mbn4#12d6fbc88b29d451_digest_top> Also, typo in the previous: XMPP instead of XMMP. Additionally, Thunderbird/Evolution Mail clients could be included too. I am replacing 802.11s with Ronja to consider health issues. All parts should be modular, anyways. Platform compatibilities is of course the issue to discuss. http://p2pfoundation.ning.com/forum/topics/ronja-open-source-lineofsight?xg_... My idea is to follow the evolution of the widest fields of the internet, for some of the latest technologies and to map them to combine it for new functions (See attachment called p2p convergence- on the right panel). The idea is inspired by Carl Woese's term, "innovation sharing protocols" http://arxiv.org/abs/q-bio/0605036 and his idea of a "denim fence" "model" of evolution as in the attached illustration Two of the latest developments in the fields of HDD storage capacity and microprocessors of last year are the advancing beyond the 2.19Terabyte limit: http://www.extremetech.com/article2/0,2845,2373917,00.asp (though using 64-bit OSs, GPT, and UEFI are part of newer platform requirements that do this by default) And and recent developments in the Einstein-Bose condensate: http://science.slashdot.org/story/10/11/26/2059208/German-Scientists-Create-Bose-Einstein-Condensate-Using-Photons(which<http://science.slashdot.org/story/10/11/26/2059208/German-Scientists-Create-Bose-Einstein-Condensate-Using-Photons%28which> may extend Moore's Law further into the future), then it is possible to combine these technologies to do much more as a client than a server would typically do. A third example of this is less of a breakthrough, but a more applicable technology and rather a sequential development in consumer microprocessing capabilities, is AMD's 16 core Bulldozer and Intel's SandyBridge 16threaded 8core chips being released in 2011. So, some of the obstacles here are both ideological and technological. Some technologies available can reproduce the internet as Fidonet, but the needs/wants of today's users either depend on or prefer heavy bandwidth for streaming/intensive processing needs. Some p2p software, as Simon at computerworldUk says, "The field is immature, but there are exciting experiments in progress." Looking at the bottlenecks of current technologies and documenting what is needed for the "minimal infrastructure" is an exciting forefront. By combining some of these technologies, it may be possible to address problems such as Wikipedia's storage needs (mentioned recently, if someone can provide a link?), and having p2p versions of those servers. The same could follow for Sourceforge.net. A theory I'm considering here is, if the some data needs of Wikipedia or other large site such as Amazon/online store increase linearly, and certain breakthroughs in harddrive capacity such as the link above extend consumer storage capabilities exponentially (e.g Multi-layer cells/3D disk writing), then as some next gen technologies physically "shrink", it may be possible to transmit enough internet application data through 1Gb or 10Gbps^n fibreoptics wires, or wireless mesh protocols such as Ronja. If tomorrow's off-the-shelf computers can handle the needs of today's mega server farms, it's possible that the exponential increases in multi-core computers with enough storage will be able to manage the needs of a real p2p infrastructure's most commonly used web applications. Trends in server architecture include low-power, high density processors: 512 Intel atom chips running at less than 2Kwh: http://www.anandtech.com/print/3768/ And potentially, ARM microprocessors applied to this technique, or Intel Chips that run at 1-100mw per core. and are stacked similarly to SeaMicro's setup. Thus, in the future, converging these trends, as an estimate, it's possible a consumer desktop would run at less than 1 watt- use 48 cores, using 2mw each, and can handle massively parallel p2p internet storage compression and decompression (or RAW uncompressed data functions) with up to 18 exabytes of storage capacity using 64-bit processors and the accompanying RAM possible (16 Exabytes). But in the near-term, it may be possible to do that with: Using an 8 or 16 core Zambezi/Bulldozer chip would allow faster decompression and compressions of .rar/.zip ,etc files of p2p Wikipedia discussion/page edits and transmitted across networks, addressing bottlenecks of bandwidth however possible. This isn't an advertisement- just an academic exercise in alternative network infrastructure theories. Giovanni On Sun, Jan 9, 2011 at 2:39 PM, Giovanni Lostumbo < -- P2P Foundation: http://p2pfoundation.net - http://blog.p2pfoundation.net Connect: http://p2pfoundation.ning.com; Discuss: http://listcultures.org/mailman/listinfo/p2presearch_listcultures.org Updates: http://del.icio.us/mbauwens; http://friendfeed.com/mbauwens; http://twitter.com/mbauwens; http://www.facebook.com/mbauwens Think tank: http://www.asianforesightinstitute.org/index.php/eng/The-AFI