Discuss mailing list

[Coyo <coyo AT darkdna.net>]

On 04/13/2013 12:39 AM, Bjorn Peeters wrote:

I will be keeping a -much interested- eye on this to see where it ends up :)
The technical side, i'm not doubting.
The financial side, if you've got - or can get - a shitload of money and build up the services slowly enough,
and can market it as a real company and not a "will fail within the first year" company (seen enough come and go),
then i would indeed assume it can be done.

But the commercial world is almost like a warzone, and there will be more than enough who'd rather not see this happen...
And again, though a law may be unjust it doesn't mean they won't shoot you down for not obeying by it.
(Though, i assume you *could* get around current implementations of these laws, it again will cost you a shitload)

And last but least, i'm not sure how it is over there in texas, but over here you'd have a seriously hard time getting the fiber in the ground,
with all the required permits and time until approval etc.

Still, someone with the money, know-how, and who doesnt like the current system, could probably pull it off.
Which is why, how unlikely it is, i'd still be very much interested to see this come to life.
And, ofcourse, Opennic would be used as the defacto dns i suppose? :)

To finish off, i'll give you a pricetag for belgium's mobile market operators:

2g license: 31 million,
3g license: 140 million
4g license: (one 45mhz block) 22 million
199 million euro LICENSING cost - i.e.: "now you are allowed to start requesting antennas and building permits".


On 04/13/2013 02:55 AM, Alex M (Coyo) wrote:

I am in the process of conducting more detailed and extensive feasibility studies, and am contacting sales representatives for the cross-connection contracts, peering agreements, transit SLAs, real estate purchases, easement and right-of-way contracts with the state of Texas in order to get a clearer idea of precisely how much this is going to cost me, and how I might meet or exceed these capital expenditures.

I'm deadly confident it can be done, but I guess we'll see? :D

Here's an update on this.

I've changed the working name for this.

My immediate friends have urged me to think larger, more ambitious, while not neglecting the immediate short-term goals to get this baby off the ground.

I need to keep the initial plans and tactics close to me chest for now, but I think you lot will be impressed.

I've broken this system up into 24 distinct parts.

The first 6 are pretty mundane, standard and battle-tested methods to keep the common carriers at bay, enforcing honesty, deflecting price gouging and demolishing high barriers to entry for small, independent or family-owned ISPs.

1) Metro Ethernet in every major metropolis.

Despite being in close physical proximity, many buildings of residents and businesses often have to tight tooth and nail to get the bandwidth, quality and affordability to Internet connectivity, both VPNs to connect to branch offices and headquarters located elsewhere, but also to each other, and to the Internet eXchange Points that even the playing field.

To change this disadvantage, metropolitan fiber optic networks between highrise condominiums, apartment complexes, office buildings and Internet eXchange Points help to keep carriers honest. This tilts the balance of power in our favor, and help to prevent the abuses by common carriers against a captive market.

2) State-of-the-art carrier-independent data centers, with Internet eXchange Points cross-connected to all other major IXPs.

Hosting providers, namely, datacenters with managed web and email hosting, VPSs and the thousands of server-related things many, many people take for granted are usually just as trapped by common carriers as the residents and small business owners who want to connect.

To change this, modern data centers capable of high-memory Virtual Private Servers, wire-speed VPN protection, fast SSDs, RAM caching, and long-term storage with affordable Digital Audio Tape drives, is critical to give anyone who needs hosting an affordable alternative, tilting the balance of power in our favor.

3) Regional Inter-ISP fiber optic networks, to bridge the Digital Divide.

Rural regions are often locked out of the bright new world of ultra-fast broadband. They are isolated from the high-definition video streaming in realtime, extremely affordable telephony, free or near-free education and training, chatting freely 24/7, all the things we take for granted. This causes a Digital Divide, which gives common carriers power over us by way of a Divide and Conquer tactic.

To change this, regional networks plug the gap between long-haul links that treat rural regions as "flyover country," and the inter-connectedness of inner-city metropolises. This gives small local ISPs a fighting chance against industry lock-out, all too often inflicted on us by the usual suspects, the common carriers.

4) International carrier network cooperatives.

Even with many local, independent, family-owned or cooperative ISPs collaborating to build regional networks, regional networks have no easy way of connecting with others of its own kind across the continent.

To change this, regional networks, and all the small and local ISPs behind them, must band together, and build our own backbone carrier networks, totally independent from the oligopoly imposed by national and international common carriers.

5) International cooperately-owned submarine cables.

Even if small, local, family-owned or cooperative ISPs manage to connect across entire continents by peering their own regional optical networks with others cross the nation, domestically and abroad, there is yet another hurdle, the submarine cables.

To bridge this watery divide, we need to construct our own submarine cables. To be as competative as possible, we need to think differently about submarine links. A step in the right direction is to use active switching and shorter cable lengths to make this a lot more affordable, and provides long-term competitiveness. To do this, we need Points-of-Presence constructed on man-made islands along oceanic ridges, well within international waters, where Maritime Law keeps any individual country at bay.

finally,

6) Affordable alternative to expensive broadband satellites.

Many rural and off-short residents and businesses find it impractical to get broadband, because they often have three choices: dialup, satellite broadband, and sometimes fixed wireless broadband. When fiber-to-the-home or fiber-to-the-premises is either impractal to deploy, has not yet been deployed, or exorbitantly expensive, the choices are positively dismal.

To change this, thermal dirigibles, hot air balloons that can be steered like a ship, sometimes called airships, are a very affordable alternative to trying to pay SpaceX to deploy communications satellites. There are many unlicensed or unused spectrum available for whoever wants to use it, and dirigibles are fully capable of such high altitude, it technically counts as low Earth orbit.

Dirigibles and other lighter-than-air aircraft are different from heavier-than-air aircraft in that they become more effecient and capable the larger they are. Helicopters and fixed-wing aircraft such as fighters and jumbo jets have hard physical limits that airships do not have.

There are a few more tactics, strategies, techniques and methods to ensure that a homebrew fiber optic broadband ISP is highly competative with the incumbent common carriers, the least of which might be called excessive and overkill.

My motto? There is no kill like overkill.

Most consumer-grade hardware, especially network interface cards and riser card buses are dreadfully slow and poorly designed and manufactured.

Purchasing a carrier-grade optical line card is often impractical for any number of reasons, the most prominent of which are sheer cost, and the fact that common consumer-grade hardware wouldn't know what to do with it, much less be able to use a tiny fraction of the true capacity.

However, as consumer-grade hardware slowly catches up to carrier-grade and military-grade hardware and software, a number of possibilities open up.

7) Thunderbolt ports.

Thunderbolt interfaces are directly connected to the processors and system RAM, which makes them as fast as the processor and RAM can physically provide.

By leveraging this potential, we can easily provide dramatically higher speeds and reliability that would otherwise be impossible for decades. Thunderbolt optical fiber interface cards would let you crank your desktop computer all of the way to 40 Gigabit Ethernet.

This eliminates the bottleneck in the silly consumer-grade home gateway routers, since you can just get a thunderbolt-equipped ordinary desktop computer, slap linux on it, enable routing and iptables NAT, and you are ready to go.

8) Wireless N+AC w/ MIMO

Wireless AC (IEEE 802.11ac) is sometimes called "Gigabit Wifi," due to the fact that it can significantly exceed gigabit-per-second bandwidth under slight electromagnetic attenuation and moisture scattering.

Combine this with Wireless N as a backup, and you have a two-punch to provide unbeatable bandwidth to mobile devices such as smartphones, music players and game consoles.

9) Unlicensed WiMAX 2

WiMAX, or WiMAX 1, is not terribly impressive. It is called 4G, same as LTE, but neither are true 4G, as 4G is strictly defined, from a technical standpoint, as greater than 1 gigabit down, 100 megabits up, nominal (not peak) bandwidth. Both WiMAX 2 and LTE Advanced are capable of this.

WiMAX stands out because there is significant spectrum left available to local communities, neighborhoods and small towns to use for whatever they feel is prudent, while LTE spectrum is decisively dominated by the common carriers.

WiMAX 2 base stations are not that affordable for individuals or small ISPs, but with the collective buying and negotiation power of entire regions' worth of small and cooperative ISPs, pooling resources like this could make the base stations a lot more reasonable.

10) Mutual peering and roaming agreements

By peering mutually with other small ISPs, and only nominally purchasing transit from regional networks and national carrier networks by way of membership dues, it elevates the technical and political importance and value of negotiation and mutual peering considerably, further tilting the scales in our favor.

11) Using treaties and neutral ground in our favor

By peering, as much as possible, in contested, neutral or international areas, such as international waters, across embassies and national borders in creative and unexpected ways, bridging borders and physical barriers as much as possible without the common carriers or their lobbyists in washington being given a chance to frustrate and thwart you, is a helpful strategy to ensure success.

12) Empowering local, municipal and regional governments

Helping small local governments, city governments, and smaller regional governments to fight off lobbyist influence from above, and helping to improve your nieghborhoods and municipalities to compete successfully. The greatest commodity local and provincial governments are families, small businesses, skilled workers and other valuable assets to move in, breathing life, increasing importance and bringing money into the local communities.

13) Brew up the perfect storm

What do you get when you combine Open Source Networking, a proliferation of Community Ad-Hoc Wifi networks, a Pirate+Tea+Green+Libertarian Party Alliance, Dramatically increased usage of Bitcoin+Tor, more Municipal Broadband initiatives and Broadband Cooperatives to supplement existing Power and Telephone co-ops? I have no idea, but I know it will be awesome.

14) Potlucks, Barbeques, Faires and Conventions

To dramatically increase awareness of Internet-related issues and concerns, it always helps to invite nieghbors and local government officials to events with lots of free pizza, cake and beer. Also cookies.

Having everyone gather around, bringing their own dishes and cookies and such potluck style, having a neighborhood barbeque, breaking out the grills and mills, hosting a fundraiser faire, or a full-scale convention at a major hotel or convention center, can go an awful long way toward bringing a critical mass of ordinary people to paying attention to terrifying privacy and security realities.

15) Distribution and redistribution of trust

Much of the Internet's underlying infrastructure is broken by design, or created with nothing we are familiar with in this day and age in mind.

Much of the Internet's underlying infrastructure we take for granted. An example of this is the centralized namespace of DNS, deliberate weaknesses to exploit against honest and well-meaning citizens, such as ICE domain seizure, DNS poisoning, seperation of SSL/TLS and DNS, lack of DNS encryption and authentication, both with authoritative nameservers and recursors.

A few other prominent exampls is the perversion of IP's end-to-end architecture, and the primary resistance to transition to IPv6 is the deliberate withdrawal of distinct dedicated IP addressing to every citizen on the planet. IP was not made with Port Address Translation in mind, nor was it part of IP to centralize all incoming connections to tightly controlled data centers well outside of the privacy and security of your own home.

Conveniently forcing citizens to relay and store sensitive information in a computer you do not own, circumventing the protection of private property laws, located in datacenters you do not own, circumventing the protection of judicial oversight and the restrictions against unwarranted search and seizure, is vital to the hegemony of the common carriers and their institutional investors and stockholders.

The reliance and dependence of every citizen depend on, among other things, withdrawing from the drug that are the pillars and bastions of the common carriers, certificate authority cartel, media conglomerates, and all manner of nasty and vicious sorts, is an imperative, if our thoroughly-earned rights and liberties.

Disentangling the nameservers from the root nameserver cartel is an absolute imperative. We should use something uncontrollable, uninterceptable, unblockable and uncensorable.

A possibility would be using an alternative blockchain of namecoin for the TLDs, and kept stable and consistent with a large contingent of authoritative and recursive 2nd teir nameservers constantly mining.

A neat way to turn the tables on ICANN would be to relegate all ICANN DNS TLDs to a preset TLD, .icann. Greasemonkey scripts or certain kinds of subscriber-side transparent proxies could rewrite links to refer to ICANN TLDs with the .icann suffix.

For example, www.google.com would become www.google.com.icann, and irc.freenode.org would become irc.freenode.org.icann.

This way, the alternative namecoin and all of the nameservers that handle 2nd and 3rd level domains are not bound to any single entity, agency, authority or country.

We might make it mandatory to exclusively use DANE, full DNSsec and DNSCurve support, mandatory TLS 1.2+ AES-512-GCM 8192-bit RSA certificates, signed by four (4) trusted DANE nameservers in different countries, on four (4) different continents.

We might make highly effecient caching and certificate validation the norm, as well as using private information retrieval to retrieve valid DANE certificates and records.

We might use a kademlia DHT to act as a backup in case of recursor or authority failure, maintaining signed records in DHT while relying primarily on DHT recursors for effeciency and speed.

Another possibility would be to use a push-pull system between recursors and authoritative nameservers, permitting much longer TTL, as any changes to records can be announced to all recursors that are subscribed to those records.

This subscription ideally would be an IP multicast address, though alternatively, it could be as simple as a P2P form of XMPP presence.

16) Adoption of P2P and F2F technologies and approaches from the infrastructural carrier side, such as passively observing Bittorrent traffic, and quickly caching the observed traffic, and seeding these files transparently. By caching and seeding bittorrent traffic proactively, it reduces strain on upstream peering and submarine links.

Carrier-side offloading could make software repositories such as apt, portage and pacman capable of using signed bittorrent distribution, with the http and ftp mirrors as a backup, entirely possible and feasible, since the carriers proactively offload strain on congested international connections, seeding as close to the subscriber as possible.

17) Comprehensive use of modular containerized data centers

Containerized data centers, sometimes called modular data centers, are small but elaborate portable data centers that are prefabricated in a factory, isolated, suspended and ruggedized in a standard ISO shipping container, with its own regulated power supply, backup batteries, a small backup diesel generator, and everything you'd expect.

Containerized data centers like this allow you to deploy important services, such as DNS, DHCP, Kerberos and SNMP, as physically proximal to your subscribers as possible, providing experience-critical components and services as quickly and effeciently as possible.

Proactive precaching, active mirroring, content distribution and software defined networking seamlessly deployed as physically close to subscribers as possible, often right down the street, can go a very long way toward reducing strain on limited network resources.

This can also permit such revolutionary concepts as transparent migration of cloud compute and storage solutions where it is needed the most, where the users are.

The bandwidth required to be able to dynamically flow network resources and services across geographic distances, without service interruption, to seamlessly and transparently migrate active TCP connections and UDP sessions across regions to process requests and synchronize information with the local greenfield data center is substantial, but not impossible, or even infeasible.

The Plan 9 Filesystem protocol, previously called the Styx protocol, showed that it is entirely possible to port existing applications without modification across network and server resources.

If programmers embrace actor-oriented programming, breaking applications down into self-contained threads that do not necessarily be on the same core, same machine, or even the same network, server-side applications could seamlessly scale dynamically across geographically diverse server farms as demand rises and falls.

The server-side logic, active processes, their execution and active network connections could easily be flowed to the very doorstep of every user that pulls information from the logic.

With OpenFlow, SDN, inter-cloud exchanges, strong deployment of greenfield containerized data centers, and distribution of critical network components, protocols and infrastructure, this sort of thing would not only be possible, but expected without question.

I think this email has already gotten a bit too verbose, so I'll leave it at 17.

If you read this far, thank you for your time and patience. ^_^