Dear Sivasubramanian,
an alternative link to the REMIX paper:
http://dl.acm.org/authorize?N19232
Dear Leandro,
On Fri, Aug 17, 2018, 4:10 PM Leandro Navarro <leandro@pangea.org> wrote:
Some replies inline.
On 16/8/18 20:59, Sivasubramanian M wrote:
On Thu, Aug 16, 2018 at 11:41 PM Leandro Navarro <leandro@pangea.org> wrote:
Hi !
Yes, the commons model is superior. CNs need to learn how to make their implementations work on a millions scale.
At B4RN they say they give the best connectivity in the UK because they dedicate all their resources to that, and they don't have to pay others.
That is what is needed to build an independant community network. Basing a Community Network on a 100 Mbps or a GBPS or two from an ISP or a Telecom company makes it a good start, but the Network is an effect a sub-distribution operation for a Telecom / ISP in terms of its dependance.I agree, there are communities that have started like that. Sharing a transit connection is a good way to reduce the entry barrier. One good case is RemIX in Scotland: http://www.cs.stir.ac.uk/~mmf/res/pubs/gaia16_remix.pdf
The above link is broken, and also the link to remix architecture from the page https://ec.europa.eu/digital-single-market/en/news/good-practice-remix-distributed-internet-exchange-remote-and-rural-networks-scotlandSeveral community networks, as they grow, aggregate more traffic, buy more capacity from different carriers, get connected to IXPs (e.g. Ninux, Freifunk, guifi, B4RN ... cases we have studied in Europe) which allow them to exchange traffic with many other networks, which results in better quality traffic for everyone involved.
In the case of guifi.net they also rent optical regional circuits thanks to its cost-sharing model to aggregate capacity (the sub-distribute/aggregate traffic you mention) from multiple individual or local participating retail ISPs that are part of guifi.net. Therefore they bridge the gap from retail capacity required by participants to wholesale open-access fiber networks.
All these are good examples to follow. The goals of good community networks such as guifi.net's goal of "promoting advanced research in the filed of open networks and infrastructures, and that of generating collaboration platforms between stakeholders" worth expanded attention. There could be a concerted effort for advanced research, and there could be a globally unified platform for collaboration on Community Networks. The Internet Society could help build one, using voluteer effort and open source technologies.
There have been and are many efforts that fit with that AIM, and the Internet Society, together other organizations, are doing coordinated efforts in that direction:
ISOC with many Beyond the Net projects, https://www.internetsociety.org/beyond-the-net. Supported by the European Commission, we are involved in the ongoing NetCommons.eu research project, the past Confine-project.eu and Clommunity-project.eu. There is this DC3 dynamic coalition, the APC Local access (https://www.apc.org/en/project/local-access-networks-can-unconnected-connect-themselves) projects, the upcoming GISWATCH issue on Community networks, the GAIA IRTF WG, the Community Networks SIG http://cnsig.info/about/ and perhaps I'm missing some.
There are also many meetings and opportunities for coordination:
GAIA in IETF locations, http://cnxapac.org/ in Asia-Pacific, DC3
meetings at IGF events, the IEEE Internet Initiative with a CN WG
and meetings https://internetinitiative.ieee.org/ among other
regional and global meetings ...
I'd say this is the role of global carriers and top tier ISPs that offer these rates in many points-of-presence globally.Just to understand, do submarine networks offer commercial arrangements for transport on a minimal scale ( 1GB / 10 GBPS )?Submarine networks that are at a higher scale of aggregation.
It would be helpful to understand what this 'higher scale' is (For example, See-Me-We 4 has a design capacity of 4.6 TBPS / s . Would it be too insignificant for this network to provide a strand of 100 GBPS/s ? for we will eventually get there, and surpass this scale (one or two new Community Networks with higher aspirations could even start at the beach front)
Even more simple, a CN with symmetric traffic (viewers but also
hosting content with demand) may get Internet access nearly for
free in a IX (the case of Freifunk in Berlin). No need to have a
dedicated undersea cable, which is "only" a point-to-point link
... Some rough numbers to understand the scale, guifi.net with
about 0.1 million users, a significant part with 1 Gbps fibre at
home does not need probably more than 20 Gbps of Internet access
(given traffic aggregation, and a big part of traffic is directly
served at an IX from content providers: CDNs from Akamai, Google,
FB, Netflix, etc)
Kind regards, Leandro.
What is the typical cost of a 1GB / 10 GBPS switch at an IXP? Do IXPs everywhere offer supernodes to Community Networks?CNs can participate in IXPs just like any other member org. The price of ports in IXP I understand depend on the size and specific infra costs of that IXP (size is really diverse). Many IXPs are also distributed (PoP in multiple locations). These are from our local IXP: http://www.catnix.net/en/taxes/
In addition, the transport/peering costs in the IXP vary depending on the symmetry of traffic. For instance, I remember talking to Freifunk in Berlin 1-2 years ago, and they were paying net 0 for traffic (without entering into details) as they have a very symmetric traffic (a good balance between content they deliver and they request, the so called "eyeballs").
Would be helpful if any of the submarine networks or its partners and IXPs share their pricing roughly.
Many of these prices can be collected or guessed from diverse sources. If you investigate that it would be useful to share the results.
Kind regards, Leandro.
Sivasubramanian M
_______________________________________________They're not the only ones. In https://www.measurementlab.net/publications/2015-Braem-et-al.pdf with independent data from M-Lab, you can see that 3 CNs are among the best operators in their countries in quality (e.g. section 4.2 pg 5, Figure 10 pg 6). We say: "The three networks are among the top eight ISPs in download speed. guifi.net is ranked first in Spain both in median upload speed and best median latency; Ninux (FusoLab) is ranked second in upload, and fourth in best latency; AWMN (part of LANCOM) is first in upload speed, 8th in best latency. In the area of Barcelona, where guifi.net has its connections to Internet carriers, the results are excellent: first in upload speed (guifi.net 7.82 Mbps, the Academic network 4.23 and Cableuropa ONO 3.31), third in download speed (Cableuropa-ONO 18.1 Mbps, the Academic network 9.8, guifi.net 9.79) and first in best latency (guifi.net 14 ms, Vodafone 25, Cableuropa-ONO 35)".
That is something known in the world of free software. There are free implementations that are superior to commercial ones (Android is based on GNU/Linux, let's see when Movistar is based on the infrastructure of guifi.net...). All commercial software products are built from free software, built openly and cooperatively (as commons), because that seems to be the only way to have robust enough software infrastructures/libraries at a reasonable cost/quality to build stable products on top.
Cheers, Leandro.
On 16/8/18 17:27, Sarbani Banerjee Belur wrote:
Hi Sivasubramanian, Community networks does exist in India and it is a sustainable initiative. Gram Marg at IIT Bombay has set up one spanning 10 villages in Palghar, Maharashtra, Digital Empowerment Foundation has set up some as well. There are more going to be set up in this year and the next. These CNs are set up in locations that have no mobile connectivity and are usually in remote, rural villages of India. Local ISPs have come to the rescue and provide bandwidth in such locations. With regards, SarbaniIn Chennai, India, I spoke to someone in an educational institution about starting a Community Network. He argued that access is no longer a problem as Telecom companies offer 3G and 4G services everywhere. He wouldn't listen to arguments concerning the cost and clever pricing models of access that indiscernably amass huge sums by microscopic extraction, wouldn't listen to arguments about nominal and actual bandwidth. He and some others take the position that a case does not exist for Community Networks here. Happens to be an iconic opinion. It is a challenge to present arguments, articles such as this are of ample help. Sivasubramanian M On Thu, Aug 16, 2018, 5:20 PM Marco Zennaro <mzennaro@ictp.it> wrote:Interesting news: https://motherboard.vice.com/en_us/article/ne5k5m/consumer-reports-broadband-company-ratings Cheers, Marco Marco Zennaro, PhD // Research Officer // T/ICT4D Lab // ICTP // wireless.ictp.it _______________________________________________ DC3 mailing list DC3@listas.altermundi.net https://listas.altermundi.net/mailman/listinfo/dc3_______________________________________________ DC3 mailing list DC3@listas.altermundi.net https://listas.altermundi.net/mailman/listinfo/dc3
DC3 mailing list
DC3@listas.altermundi.net
https://listas.altermundi.net/mailman/listinfo/dc3
--
_______________________________________________ DC3 mailing list DC3@listas.altermundi.net https://listas.altermundi.net/mailman/listinfo/dc3