Hacker News new | past | comments | ask | show | jobs | submit login
NYC Mesh – community-owned network to replace your current internet connection (nycmesh.net)
550 points by tonyztan 6 months ago | hide | past | web | favorite | 170 comments



I run a website about starting and running small wireless ISPs here: https://startyourownisp.com

We also have a matrix chat room here: https://riot.im/app/#/room/#startyourownisp:matrix.org

Don't want to detract from NYC Mesh or the conversation at all just thought this could be interesting for some of the folks on the thread.


I have contemplated doing this for many years. I probably won't, but I am really excited about learning. Thanks!


Thanks for this, we are deploying our WISP in the next month or two.


I looked heavily into doing this about 8 years ago in a rural town in Australia

Sadly the "National Broadband Network" would've rendered it obsolete by now


There's a similar effort gaining traction in the Los Angeles area: https://www.communitybroadband.la/

I've also noticed that local municipalities are getting into the ISP business. Beverly Hills, for example, is currently deploying a city-owned/operated fiber-backed ISP: http://www.beverlyhills.org/fiber


Josh from the Community Broadband Project here. Would be happy to answer any questions about our mission and progress in LA. I'm no regular, but we just got a bunch of traffic from Hacker News and I'm here to spread the gospel.

NYC Mesh is doing very cool stuff--we wish them the best. We're in touch with a few other community ISPs who are growing organically in competition against big incumbent ISPs. It's inspiring.

Beverly Hills, West Hollywood, and Culver City are all constructing municipal fiber right now. In the case of at least WeHo and Culver, it's aimed primarily at businesses. Some residents may see fiber offered by the incumbents, but we did some math of those offerings and it was $133/mo for 1Gbps. It'd be more helpful for the community if reasonable speeds were more affordable and available, than only some wealthy neighborhoods getting fiber.


I don't know if laying cable is one of your biggest challenges, but I assume it would be along with right-of-way.

What do you think about the idea of using the sewer system to lay either fiber or 10+gb copper? Before you think about how gross that is, consider the advancements in drone robotics. You would need to devise a cable-laying robot that could traverse the sewer via remote control. Mount fixtures to the top of the pipes and string the cables through, connecting neighborhoods.

This would require convincing the city to lease you the right-of-way and put up a bond to cover any repairs to damage caused by the cables. But it would solve both the bandwidth problem and the "those are our telephone poles!" Problem. What do you think?


We're using fixed wireless. Advances in that technology have made high speed broadband feasible in areas with high wireless interference like LA. We believe wireless is the best path for what we're doing. It doesn't require interfacing with existing infrastructure anywhere but at the towers/APs, it deploys and redeploys quickly, and can provide competitive speeds and beyond.

More importantly, it gives us the freedom of building our own infrastructure, since access to public infrastructure is prohibitively expensive--by design, it seems.


Water utilities are (justifiably) very protective of their infrastructure. Justifiable because of the extreme expense in closing a street, excavating, and repairing anything that happens to their buried infrastructure. I've seen the one in my city reject a stormwater pipe because of a couple small pieces of crusher dust. I can't see them being positive towards the deliberate installation of another biofilm-forming-surface inside their pipes. But YMMV I'm only speaking from experience in my small neck of the woods!


I don't disagree. I think that a full mitigation plan could help. For example showing that fiber can be easily chopped up if a pipe clogged. Maybe biofilm could be mitigated with an electric charge on a conductive outer sheath? I can't see how a single cable or 2 would ever require digging into the street.


Even a single cable in the sewer can require digging up the street if the cable causes an obstruction in the pipe.


Rule #1: Never install new copper in the outside plan.

That being said, there are companies specializing in installing fiber cables into sewers. They use tracked robots to install steel bands inside the pipes that keep the fiber cable in place. It's not cheap, but it is cheaper than ripping open the streets in a city.

There are also techniques to install fiber cable into water pipes.


You might be the totally wrong person to ask this and I apologize if this is a naive question.

Is there a way a lay person (non network-engineer) can somehow get involved with their community to help get this started? I'd love to see something like this get started where I live (Minneapolis) and would be willing to help in any capacity I could.


Like any business endeavor, if you don't have the skills to do it yourself, you need the resources to gain the skills, either through learning, networking and meeting people, or hiring someone. The scale you start at changes things too. We've spoken to a few grassroots ISPs that started 3-5 years ago by non-IT folks who just wanted to share with a few neighbors. Over the course of time they learned and grew the network and now they're at a couple hundred customers. Not big in the ISP sense, but it provided an essential service to their community and competes with [and steals customers from] incompetent, or at worse, anti-consumer incumbent ISPs.

If you have the passion, but not the IT skill set, be the banner waver. Be the leader of the endeavor. Use your passion to convert others. You'll find your talent.


One way that a lay person can help is with rooftop site acquisition, negotiation, contracts, marketing and presentation materials. Or as labor to install stuff, if it is a real nonprofit. Own a van that you can tie a 18' extension ladder to the roof? Great, you get to be the guy buying 16" concrete blocks from home Depot for the roof mount ballast.



Hi Josh.

Forgive my ignorance, I am honestly just curious. I took a look at your test area on your website, and it is filled with homes/condos that are in the millions of dollars. What practical reason would someone that could afford a home that costs this much need to utilize your service?

Is this area essentially exactly that, a test area, and not intended to be the location you deploy in first?


Our intended initial deployment was in Culver City, off a tower that had reach to Culver and city district 10. We met with LA city, Culver City, and did a lot of prep in that direction. Unfortunately, the critical path to fiber was delayed in construction for a year, with another year to go. In addition, we were thrown a curve ball with a tower engineering study prior to that tower's use that would eat up a sizable portion of funding.

Right around this time we were offered funding from WeTransfer as part of their Net Neutrality campaign, but under the obligation serve their area in Venice. We decided to go for it, and use that as a launching point for a proof of concept network. While not ideal for obvious reasons, it goes a long way towards proving our tech and business model. Our hope is that we can grow out the network towards Culver and CD10, acquire some critical infrastructure there, and then push east across the West Adams/Jefferson Park corridor.

It is a true deployment--we don't plan to uproot the Venice service, but it may end up being an arm of the network as opposed to the core in the future. One of the key goals of our project is to free low income neighborhoods from stone age technology by offering broadband speeds far above DSL for a fair price.


I should add that it'll take us some time to tap into the federal programs and grants to truly make this incredible for low income neighborhoods. Venice has been serving us well as a means to get our sea legs, so to speak. Starting there won't impact our plans for closing the technology gap where it exists.


Thanks for the thoughtful reply. I appreciate it - makes sense! I wish you all the best of luck.


Hi Josh!

Do you buy any transit at all or rely solely on peering?

Who do you peer with, and how easy has it been to find peers, due to the huge asymmetry in traffic?

Do you have any CDN caches deployed (GGC, Netflix, etc.)?


I'm not related, but any small ISP is going to need transit. Larger ISPs have pretty high requirements to peer, but random sites will be single homed in those networks; to say nothing of overseas sites.

But, most of the networks that provide the majority of the traffic have pretty open peering policies, and the LA area has a lot of places to peer, and peering places are also usually a reasonable place to get transit so... If you're seeing at least a few gbps from FAANG, they'll peer with you, if you're where they are, and all of them have PoPs in LA, because everyone is there (this is a little circular, LA was highly connected in the PTSN because of population size, and that lead to early internet exchanges and undersea cabling, which leads to more people wanting to be there and so on)

I'd imagine if they have the traffic to justify it, and the space to house it, they'd get the CDN caches. It's a pretty good win, peering is settlement free, but access to the peering point isn't free; keeping traffic from using your ingress capacity is nice (depending on your network architecture anyway)


Great tech questions. Right now, in our beta/test network, we're receiving DIA service to a remote server location and that's it. No peering, no CDNs. Our next steps are to increase the bandwidth of our DIA, since that's more cost effective at our current scale than we'd benefit by peering. The step after that is to swap that DIA for transport and then colocate in DTLA to find peers. CDN caches often require a minimum number of customers on network, something like 5000 for Netflix, if I recall, so we'll have to grow before we can take advantage of caching.


A Netflix cache appliance isn't dependent on the number of customers, it's on dependent on the amount of traffic from Netflix. You don't need 5000 Netflix customers, you need 5 Gbps of peak traffic to qualify.

However that isn't your only option. You can just peer with Netflix, they have an open peering policy.


The issue with peering with Netflix is that Netflix only peers in certain locations, so you need to buy transport to the exchange where Netflix is present.


Hi, is your product based on the Necto offering?

https://apply.nectolab.io


No, but Necto looks like a very interesting way to make ISPs more approachable for a community. That's one of our own goals, but in our case we're doing everything in house, which makes the most sense given our expertise.


CBC has a very good piece on NYC Mesh: http://www.cbc.ca/news/technology/wifi-nyc-mesh-new-york-cit...

Also André Staltz's TEDx Talk that mentions it: https://youtu.be/UjfWAbGfPh0

I really do hope this succeeds and takes off amid the likely scenario that SCOTUS will eventually kill Net Neutrality for good in the next two years. Things like this are the way to go if we want a bit more freedom in terms of surfing the web.


You may enjoy: https://staltz.com/a-plan-to-rescue-the-web-from-the-interne...

Also, if you're curious about what that looks like today: https://scuttlebutt.nz


Come by Sudo Mesh meetings at Omni Commons in Oakland on Tuesdays at 730 for https://peoplesopen.net/

Or keep an eye on major repos like the firmware: https://github.com/sudomesh/sudowrt-firmware

There are many places to help. Yes, writing code, but also hands-on stuff like mounting radios on peoples' roofs, grant writing, documentation, etc. Even things like establishing consistency across projects: https://github.com/sudomesh/bugs/issues/34

I've really enjoyed it so far this year as a way of practicing software production skills with other community members and non just inside the context of a for-profit startup.

A lot of pieces are far enough it would also make sense to set up meetups and meshes in SF and Peninsula as well.

Tuesdays. 730pm. Omni Common near Macarthur BART stop in Oakland. Come help!


What about a mesh that is not connected to the internet?

These days, people use the internet like TV or radio: many consume, few produce.

What about a "community" mesh, where e.g. a small town has websites, email, webservices, gaming hosts etc for shops, businesses, sporting, clubs, council, local news and social media for that town?


https://disaster.radio is a neat little project that might be helpful for building something like this.


I've been working a bit on https://baculus.co which has some overlap with that use case. Super early hacking, mainly built on scuttlebutt and other services on rpi3s, all in backpacks.


It's entirely possible to build a WISP and have no transit upstreams or peers. If you wanted to truly island it you'd need additional DNS infrastructure that most wisps don't have. Never seen it done, though, because the use cases for such a thing are rare, vs a wisp that provides a default route to its customers out to the internet.


For the same reason communities don't build a second road system not connected to the main road system. Why would I want to build a system that makes it so I can't check websites, email, webservices, gaming hosts, businesses, clubs, council, local news, and social media just because I'm in the next town over during work?

Also if you really wanted to only have services available locally why not just do that via access control policy instead of building a second infrastructure?


That place still exists on the internet. The 90s are alive on Tor.


can you elaborate on this? would love to explore, but where exactly to start?


Check out Broadband-HamNet.


>>> a small town has websites, email, webservices, gaming hosts etc for shops, businesses, sporting, clubs, council, local news and social media for that town?

Facebook.


So I have been toying with spooling one of these up for the past 6 months. I live too far uptown to get line of sight for their supernode, but I do have not 1, but three LinkNYC kiosks on my corner that offer gigabit for free. Wanted to test out using their LinkNYC option. Was going to use wireguard as a vpn backend on a router.

The only downside for me is that I need to "re-log in" to google once a month. The wife would not be amused. Plus, we are currently paying for 200/20 for about $45 from spectrum.


The re-logging in problem can be solved via script. Check for the captive portal login page. If present, simulate clicking the log in button.


Stuff like this make me long to live in NYC. Only city in the US with the density, technical expertise, and money for this kind of thing to happen.


I thought the same was true for the Bay Area, hence why I moved here. I was genuinely surprised to find Google Fiber isn't even available, let alone any mesh networks.

In all other aspects of technical expertise though it's blowing my previous city, Houston, away. Companies open to candidates based on technical expertise alone (rather than throwing non-B.S. degrees straight into the garbage), lots of great meetups, lots of fellow geeks doing exciting work.

Actually, another thing that surprised me was lack of hacker spaces - Houston seemed to have the same amount of them... one. The only major one I'm aware of here is Noisebridge, same for Houston which has TX/RX.

edit: oh, apparently SF has quite a few more than I realized. So does Houston, but not as many.

https://wiki.hackerspaces.org/San_Francisco https://wiki.hackerspaces.org/Houston


Monkeybrains (https://www.monkeybrains.net/) operates a mesh network in SF. They're not community owned but they are very much a local business.


I wanted to like Monkeybrains but they were just too subpar for the competition in terms of speed, reliability and flexibility. I switched to sonic and get 1GBps fiber for barely $12/mo more than what I paid MB.


Dunno where you are in the Bay Area, but in SF I have Sonic which is synchronous gigabit for ~$40 a month.

It's not Google Fiber, but it's close.


For some reason they offering only 10Mbps in my building in downtown SF... compared to the 150 I get through Comcast. I hate their guts, but they do at least give me pretty darn fast internet.


10Mbps sounds like Sonic Fusion (AT&T DSL partnership) and not Sonic Fiber.


I live less than 10 minutes from the Google complex and google fiber isn't an option. It's really sad.


There's definitely some mesh networking stuff going on, e.g. https://peoplesopen.net with weekly meetings at https://omnicommons.org

There are a large-scale effort in SF many years ago... but it's a hard city to build in. You can read a bit of that history here: https://en.wikipedia.org/wiki/San_Francisco_Municipal_Wirele...


I know people doing this in Detroit on a smaller scale. The nodes are I believe using cable modems, not 20 GB connections. But it's free because they have money from a foundation and it's all volunteers. It's all about getting the poor on the net.

https://motherboard.vice.com/en_us/article/kz3xyz/detroit-me...


Better still, if you live in Brooklyn you can actually subscribe to an independent locally-run fiber ISP: http://bkfiber.com/

It's more of a business product than a consumer product, and it's got limited range, but it's great to know the option even exists.


Unfortunately it's limited to very few places and is practically non existent if you consider Brooklyn as a whole.

Meanwhile Brooklyn has Verizon fiber gigabit network and supposedly getting one from Altice too. Two gigabit fiber ISPs in one place would be something quite unique.


NYC is a nice place to believe in in an idealistic way, but crummy when it comes to the nitty gritty function of the city.

Source: I live here


Could you maybe elaborate on some examples you have in mind?

I'm strongly considering accepting a job offer for a company in NY since I'm starting to get bored of living in Seattle. I figure moving to NY would mean being surrounded by more people and having more opportunities to collaborate on projects like these.


Oh :(


I've been wanting to get involved in this kind of thing for a while now. Shame they use Slack, and not an open platform like Matrix or IRC, though!


This is a really annoying behavior. Not every project needs or even should be perfect along every dimension that you or anyone else cares about.

Would you also find it to be a problem if they held a meeting and didn't serve fair trade coffee? Should they expel any members with politics you find objectionable? Must they demand that everyone give up their iOS and Android devices too because they're surveillance instruments and running non-free software too?

Here's a good Tumblr post this made me think of:

- [The Grey Tribe — Activism, Accessibility, Setting Good...](https://the-grey-tribe.tumblr.com/post/175038067648/activism...)


Nitpicking about tool choices is not a useful contribution to a volunteer project starved for resources. There are better ways to contribute.


there are certainly better ways to contribute, but i’m not sure it’s nitpicking. it seems like this issue is pretty close to the network’s core values and reason for existing: getting away from centralisation, censorship, big companies, etc


This reasoning is a distraction. The fact is that it is unhelpful and slack works fine.


I do not like slack either.

that being said, get involved and get them in Matrix or IRC, not too difficult to enact change I think people could get in on it with appropriate leadership!

zulip looks interesting to me as well.


I have repeatedly offered to pay for hosting a standalone Matrix server for the Julia community, and so far been ignored. Seems like people just don't care.

I think the best I can do at this point is set up my own server running Matrix (or Zulip, which I haven't really explored) and host my own bridge to ease the transition. It's the classic problem where, now that I actually have money to spend, I don't have free time or infinite energy for side-projects anymore.


As much as you're trying to help, the truth is, small projects don't have the resources to do stuff like this. Now they have to devote time (probably someone's limited after work time) to bringing you onboard, getting you setup, etc. Then, you set everything up and let's say you disappear. Now they have this custom setup that only one person knows how to operate. At the end of the day, Slack works. They're already using it. No one has to get trained on new software. You don't need new accounts and passwords and admins and teams. The list goes on. This is a big issue with small projects - everyone wants to join and change some thing that's not part of the core mission.


They used to run a Matrix bridge, but from my understanding the administrators had no interest in maintaining it.


Also,

riot.im

Mattermost

Zulip

Wire

Rocket.chat

and more here: https://alternativeto.net/software/slack/


Riot.im is a Matrix.org client.

Mattermost is nice but it's not federated and I don't think it will catch on as a "public-facing" chatroom option. It's a nice solution inside an organizations that doesn't want to pay for Slack. I haven't used Rocket.chat but it looks similar along those lines. Wire just doesn't seem like the right solution.


I was participating in NYC Mesh for awhile, but eventually I had to unplug because I got notices from my ISP about copyright infringement and I didn't want my internet access to get shut off. So sad. I really want this effort to succeed.


Were you peering? I'm under the impression that they would not use an ISP connection for individuals. Thus, you would have no ISP account on which to receive copyright infringement notices.


Freifunk, a somewhat similar German project solves this by using a mesh VPN to tunnel traffic out.


why would i go through all this effort just to let some stranger torrent iron man on my internet account


Collective action can give all of us, including ourselves, more freedom. This isn’t about torrents, it’s about the freedom to use the internet without surveillance and monitoring of our every move.


Obligatory tragedy of commons reference


The tragedy of the commons is not a natural law.


It might be kinda something like a natural law:

- [Great Oxygenation Event - Wikiwand](https://www.wikiwand.com/en/Great_Oxygenation_Event)


The general idea of Meshnets worldwide is to turn 'my internet account' into 'our internet'.

all this effort = buying a router and plugging it into your computer.


Depending on your jurisdiction all this effort can also include legal fees, visits from the police and saying good bye to all electronic equipment for a couple of months while it's being investigated.


IIRC, also being registered as an ISP, giving them certain protections.


This isnt much of an issue now that NYC Mesh is providing egress in place of an ISP. There aren't many individuals left broadcasting their own ISP connection out.

It sounds like you were sharing your own ISP connection a while ago before this was the case, mightbyte?


It's been awhile, so you could be right. I just bought the router from the NYC Mesh people and hooked it up to my internet connection thinking it would be a nice way to give access to people in need and support what I consider to be a powerful idea and a worthy cause. It was never a big enough priority for me to put effort into figuring out precisely what was going on. If all NYC Mesh traffic in and out of my ISP goes through a VPN, maybe I should give it a try again.


Can you share any more detail about this? My impression was that anyone who shared another internet connection with the mesh does it through a VPN tunnel, so traffic exits through the mesh's PoP.


Have you seen https://altheamesh.com/?


Does anyone know who NYCMesh peers with? From their FAQ:

"Are you an Internet Service Provider (ISP)? No, we are not an ISP, we are a community that shares ownership of a network. Collectively we have our own internet connection at our “supernode”. Also some of our members share their own internet connections.

Our network peers with other networks at an internet exchange point and provides access to the internet without traditional ISPs. We are a non-profit project of the New York chapter of the Internet Society (ISOC-NY)."


They publish their peering details and AS here:

https://nycmesh.net/peering/


Thanks. It's still not clear to me how MYCMesh users would be able to reach the internet.

They state in their FAQ they are not an ISP which leaves me to believe they don't buy transit from anyone.

But looking at that peering link there are only 122K prefixes being advertised at DE-CIX NY and NYCMesh only peering with 14 other ASes there.

That would leave large parts of the internet unreachable. Perhaps Hurricane Electric or one their other 14 peers at DE-CIX NY have agreed to be act as transit AS for NYCMesh?


packet.net is main transit provider, and think there are a couple alternatives now as well


Ah thats the bare metal cloud provider. They seem like a cool and interesting company too. I'm glad to hear they're involved. Cheers.


I have a few naïve questions maybe someone can answer:

1 - Why aren't they recommending any of the newer routers like the Asus RT-N66U/RT-AC68U? They're well supported by OpenWRT and more available.

2 - To start a similar service in another city, do you need fiber already to be laid? For example, I'm in Pittsburgh and we don't have any fiber anywhere, does that mean it'd be impossible to have a community mesh like this here?

3 - Is there a way to find the "internet backbone" building in your city?


I'll take a stab...

1. They are likely using point-to-point/point-to-multi-point directional wireless equipment, designed for longer distance high bandwidth wireless connections where you know the fixed location of the transmitters/receivers and can tune/optimize for that purpose. Those devices you mentioned have hardware designed for omnidirectional local WiFi (broadcast & receive everywhere within x,00 feet). You could have one of those in you're home to provide wifi, and a point to point (with a directional antenna device designed for building-to-building/house-to-house connections) on you're roof to connect to the mesh network. Think of it as a mini satellite dish, pointed to the location from which the signal is being transmitted. Also, using standardized equiptment is more practical to efficiently deploy monitor and manager for a project like this.

2. No. The fiber will be there eventually, and wireless bandwidth is by far slowest link in the chain (the theoretical max speed of 802.11ac is eight 160MHz 256-QAM channels, each of which are capable of 866.7Mbps). Cat6 ethernet cable consisting of 8 conducting wires is capable of providing 10,000Mbps (10-Gigabit) maximum data rate, up to ~100m between powered devices. There are several types of fiber, OM4 is capable of 100,000Mbps over distances up to 150m. These are theoretical speeds, IRL divide by 2 twice and that's a good bencmark target (by 5 for WiFi ;-).

3. Internet Exchange Point https://en.m.wikipedia.org/wiki/Internet_exchange_point

Usually you're looking for 'co-los' (co-locations), buildings that host multi-company/parties server and network equipment and have high bandwidth connections to internet backbone provider networks.


Pittsburgh has a wireless community network, http://www.pittmesh.net, which is run by Meta Mesh Wireless Communities, https://www.metamesh.org. I'm involved!

MMWC doesn't have its own fiber but does have access to KINBER's PennREN fiber network, a little-known gem in the commonwealth that happens to run down Fifth Ave and a few other main roads in the city. We're working on rolling out access to it. As we're a donations- and grant-funded non-profit, our rollout will be slower than a commercial operation.

Pittsburgh has a lot of fiber running through it, but most of it is commercial backbone fiber. One data center that I know of had 12 ISPs peering with it, and that was in 2009. Only Verizon FiOS has executed a residential rollout in the city, and you can read more about folks opinions on that in /r/pittsburgh.

I've passed this thread to the MMWC executive team for them to add more detail!


Thank you so much! I'd love to get involved with PittMesh in any way I could.


1 - don't have a good answer here

2 - a service like this doesn't rely on much fiber at all, outside of the point(s) where the mesh connects to the rest of the internet. Its almost all point-to-point wireless, with the points eventually converging back on the base stations at datacenters.

3 - the common term for this kind of facility is a "carrier hotel" or "ix" (like internet exchange). A carrier hotel is usually a dedicated-purpose facility where lots of different providers connect their networks together specifically for that purpose, but, hypothetically even a reasonably well-connected commercial datacenter will have at least a couple of carrier connections on site who could be used to create a mesh network base.


Hey there, I run Philly Mesh and am friendly with the MetaMesh folks in Pittsburgh. You might want to give them a shout if you’re interested in this stuff, https://www.metamesh.org


regarding point 1- im going to speculate that they are recommending hardware they are familiar with and know will work. the asus you are talking about looks like a smart router and might need to have a workaround to play well with others. ive had the hardest time trying to do anything technical with smart routers without just installing something like WRT and forgetting about the smart router stuff. im not sure about this one but it might pull a linksys and store your configuration data somewhere offsite, that would be an issue for me if thats the case as i value my privacy and local control of my hardware. if you feel comfortable working your router configuration it wouldnt hurt to try it


I think we had this before a few months ago. It is no different than any other regional transit provider (in the case of their fiber loop) or wireless ISP. You still have to pay for decent peering and you still have to deal with the laws of physics and claude shannon when deploying wireless networks.


I recall the NYCmesh people having free bandwidth from an employer or sponsor. That obviously helps. Not exactly scalable or replicable tho.


There was a donation of bandwidth for supernode 1, but the cost of that bandwidth is insignificant relative to other costs. The current model is scalable. Supernode 3 is launching in the next week or so and Supernode 4 will be following it soon.


How are you covering costs?


Cool initiative based on positive ideals. From a practical perspective, NYC citizens have a number of good options for internet access; however, perhaps learnings from this community project can be brought to more needy areas outside the US.


Do you live in NYC? I do, and I don't have at all what I'd call "a number of good options" for internet access. In my area, there's two companies who operate similar to a monopoly (Verizon and Optimum), and regularly raise prices arbitrarily and often for no reason.


Having two choices actually might put you well above average for the US. Back when I was in Mountain View, CA (a few miles from Google HQ) it was either Comcast or nothing and the irony was not lost on me.


AT&T had an offering when I lived in Mountain View that was so comically terrible as to not realistically be an option at all.


This is still true as of last year. Comcast was the only viable option.


You get two? My block on 78th and 1st has Twc, I mean Spectrum, and I don't know what I'm going to do when they get kicked out of the state.


Ah, I meant 'good' relative to options in the developing world. Getting +20Mb/s for under 0.1-5% of your annual income puts you way ahead of most people on Earth.

But yes, I'm sure there's cities out there with more options than NYC.


Yeah... every apartment of mine has had exactly one option, TWC=>Spectrum (in Manhattan) or Optimum (in Brooklyn).

I've lived in 4 apartments in 7 years and never once had a choice.


A lot of this may have to do with the building & management -- often times, the building's condo/coop board strikes an exclusivity deal with the ISPs and get a share of the revenue.

[I'm on a condo board]


I live in the West Village and have ONE choice... Spectrum :(


Internet choice really comes down to what building you live in. A building might have exactly 1 ISP available, or there may be 3+. Always ask before renting if you can't afford to be saddled with DSL-only internet access.


I love the idea but how do you connect to the greater internet? What will happen when you get a request to say what user was running on a certain day talking to whatever, or handle NSLs?


From the article:

"The network connects directly to the internet backbone, so we do not rely on an ISP."

Their peering policy: https://nycmesh.net/peering/


I don't understand what that means, "internet backbone?"

Is that a physical location somewhere? I've read on the peering link, are they saying there's an "internet ethernet port" at 375 Pearl St, New York, NY that they plugged their whole mesh network into? Where's the ethernet cable go from there?

I'm probably asking a bit too much here (how does internet work lol) but I don't understand how they've bypassed all ISPs... I just figured Comcast, being an ISP, also has a controlling stake in whatever means by which it gets data across the pacific ocean or whatever.


I work for a CDN, but am not a networking expert. I will try to answer as best I can:

Yes, it is a physical location. Often called 'an exchange', there is usually a building in major cities that most networks connect together at. You rent from the building owner, and they provide a 'drop' which is basically a cable you can connect to your core router in your part of the datacenter you are renting (your 'rack' or 'cage')

From that point, you set up BGP sessions with your 'peers', other networks you have agreements with. Sometimes those agreements are for 'transit' (which means you pay per bit sent across the wire, but you can send all your traffic that way) or they are peers you don't pay (you might have other agreements, like we only send this much traffic or that hte inbound and outbound traffic has to match)

According to that peering link from the above poster, these guys will peer with anyone. They basically want as many peers as possible, and are hoping those peers agree to route as much of the internet as possible.

The way BGP works is that each side of the BGP session tells the other 'i will accept and route traffic for these subnets', for example "i will accept packets destined to any IP address in the 1.2.3.240/24 space" The /24 is what is known as a CIDR, and is a way of writing "all the ips between 1.2.3.0 and 1.2.3.255", 256 IP addresses.

So if the core router on my side gets a packet from my network destined to 1.2.3.123, I could send that packet on to the peer who is advertising that they will accept packets destined to 1.2.3.240/24. You need to have a peer for every IP address to be fully connected to the internet. If you don't, you would normally pay for transit, which will route for all IPs.


I just wanted to say thank you for explaining this. I've read about this before, but I feel like something finally clicked when reading your comment.


Don't worry, most people don't understand the physical infrastructure of the internet. However, from my understanding, a direct connection to the internet backbone is what gives tier q ISP's their God-like monopolizing power in a given region (like Comcast in most parts of Baltimore city).

However, the backbone nodes require maintenance, so the fee to connect directly is expensive. You can't simply move next to a backbone node and plug in. I presume that backbone fee is the primary reason for the individual's fee for hosting an NYCmesh node.


> However, from my understanding, a direct connection to the internet backbone is what gives tier q ISP's their God-like monopolizing power in a given region (like Comcast in most parts of Baltimore city).

No, the reason Comcast and other incumbents dominate is that they built out the last mile infrastructure. The backbone costs are a minor expense.


Glad I could help!



This Wikipedia article should answer these kind of basic questions:

https://en.wikipedia.org/wiki/Peering


Someone is always your upstream, or you have a port at a peering fabric (and anything not on-net at the peering meetup goes through a transit provider).


A tiny AS like this certainty relies on their upstream transit ISPs. I doubt they're getting more than 1/8th of the global v4 routing table via local free peering.


Are there any similar initiatives in London?

Failing that, any Londoners who are interested in starting something like this?


Not that I know of, and I don't see a major business-case for it. The UK ISP market is much better than its US counterpart (we have no Comcast, for starters) so I'm not sure there would be enough demand for a wireless ISP.

But if you have any ideas on how to make this profitable, I'm up for it. Email in my profile.


Correct me if I'm wrong, but ain't Relish (https://www1.relish.net/) a WISP?

If you guys gather around to chat let me know, I might be interested in participating as well.


They use LTE which doesn't really scale well and their reviews have been less than stellar.


Are the organizers in this thread? I wonder if you folks were aware of cjdns, and if so why the decision not to use it? I've been thinking about setting up a cjdns-based mesh network in Philly but I don't think I have line-of-sight to any geeks.


Early (~3 years ago) NYCMesh used cjdns. I am not 100% sure of the reasons, but I think at the time cjdns had bad performance and was much harder to setup. I personally tried out cjdns recently and was pleasantly surprised with the ease of setup, though I still think hyperborea is having scaling issues and Guifi (the largest mesh) doesn't use it.

There is nothing stopping anyone on NYC Mesh to use cjdns, and some of us do run it for certain services.


Cjdns’ crypto is a known cpu bottleneck, meaning it isn’t well suited for soho routers. Some of us mesh folks are squeezing power out of SBCs with great results, but cjdns’ roadmap might have us seeing more performant operation coming up :)


What gives - this is the only mention of Guifi.net on this thread - and Guifi is the world's largest Mesh, by an order of magnitude! 56k+ nodes operational in Catalunya!


Hi, I run Philly Mesh. You might be close to some other geeks, but if not you can always use cjdns' overlay ability :).

Come chat if you’d like, https://phillymesh.net


Nice! I joined the #phillymesh channel earlier but it seems to be +m.


The bay area seems to have one such Mesh network, in Oakland: https://peoplesopen.net/

It would be great to see something like this on the Peninsula


"The network connects directly to the internet backbone, so we do not rely on an ISP."

How does it work? Can anyone connect directly to the Internet backbone?


I think that is just meant to be an approachable way of saying that NYC Mesh peers on public internet exchanges

https://nycmesh.net/peering/


Anybody with enough money can connect to one or more IP transit providers ("the Internet backbone"). It will usually set you back a few thousand dollars per month.


Here's another project:

https://www.rightmesh.io/how-it-works


With all due respect to this project, I’m sure it has a lot of great potential, however:

> Our patent pending switching technology ensures data integrity.

I personally have little interest in trading one network which is controlled by a few people for another which will just be controlled by different set of a few people. Even if it is powered by trendy blockchain or whatever.

I am incredibly excited by some of the meshnet work we are seeing, but any of these which are just seeking to position themselves as power brokers on our future communication systems really need to be left on the wayside.

We’ve already seen firsthand what happened when our first iteration was turned into a glorified shopping mall. I’d prefer that we avoid this fate with the second iteration.


Meh. I put more faith in our corporate overlords developing decent wireless internet service via 5G over NYC Mesh really taking off.


This doesn't seem to actually be a mesh network.


Yes, it is, though the homepage doesn't seem to explain it well.

You can connect to their "supernodes" which are the backbone, that connect to the Internet proper. You can also connect via other member nodes. So by my understanding of the colloquial term "mesh" for wifi, it is.

I do take issue with their FAQ stating they're not an ISP. They are providing Internet access to the public. They are an ISP, even if their structure and methods are unconventional. Maybe they're avoiding the term due to legalese?

It's an interesting project. I wonder if something like that would work in urban San Antonio... Hm!


It's not really a mesh network. It's basically stick a dish on your roof and point it at a tower downtown that has a direct connection to the internet.

Yes you can then extend it a bit a sort of create a mesh, but it's not really the same as having a mesh network which spreads out evenly from individual access points.


This is not really a mesh neteork. It may resemble a mesh in a logical diagram, but it's composed of PTP wireless, ptmp, and fiber.


Shout out to a fellow San Antonio local!


Kai from Ultimate?

internet at cyberfoxfire com


You are being downvoted, but this is a good point. People would usually expect it to be a peer-to-peer mesh system without any supernodes.

The problem is that in the experience of the similar German "freifunk" project, which tried this with custom router firmware, this really doesn't scale well, even with a custom developed routing protocol (B.A.T.M.A.N). So many Freifunk communities are also choosing this supernode setup as a compromise between "full" mesh and good connectivity.


From the article:

"All of our network nodes cooperate in the distribution of data, and the network can also function independently in case of emergencies or internet shutdown."

Sounds fairly meshy to me.


Something like bgp+ospf adjacency between routers and ring/triangle shaped networks can provide a high degree of resilience, but are not "mesh".


Agreed. Seems to be fairly static routing, albeit using BGP. Doesn't appear to be a mesh protocol in sight.


BGP is mesh, we also use bmx6


It is a hybrid at this point. Supernode and hub connections are not mesh connections, but there is some second hop meshing still happening between nodes where there is not line of sight to a hub.


Can you please elaborate?


I think I was a bit too terse, but I really feel like it's on them to explain themselves better. They link to this video. https://www.youtube.com/watch?feature=player_embedded&v=glW_...

That does not describe anything related to a mesh network. It appears that the people running it are joining nodes in a mesh network, but they don't seem to be encouraging you to join in a mesh network.


city-mesh movement started about 10 years ago, meraki etc are "hot" then, it never went away and stayed relevant, but not as trendy as it used to be.

libremesh is based off openwrt, it needs a web-based dashboard, the firmware seems solid, very interesting.


Is there any way to connect to this from the other side of the Hudson?


We have a #newjersey channel on our Slack. We need more people there to help organize


Pringles can.


not sure why this was downvoted, a Cantenna [0] was originally created with a pringles can.

[0] https://en.wikipedia.org/wiki/Cantenna


A Ubiquiti radio and parabolic dish don't cost all that much.


parabolic wok-fi



Does it comply with Net Neutrality?


this seems freaking rad.


A big problem with meshnet sofar is that it goes against the status quo, and it cut out the moneymakers and regulators, thus mesh nets have segment distances limited to ethernet and repeaters as well as where does physical infrastructure live? utility poles are owned by someone other than common people, and a trench transect for fibre is limited by property concerns...we need to develop a method of long hauling signals that wont be interfered with and will not interfere physically with anyone else...optical linkage works and has no palpable medium to be destroyed by hostile action. The bandwidth of a line of sight laser is very high, so high that it doesnt need to operate 24-7-364.25 of the year...if two or more LOS linked supernodes cache each other you win...if something gets in the way, point your LOS to a different LOS caching peer...


How robust is something like a laser in the face of weather interference, like rain or urban smog/haze?


Not very. In the (admittedly old) setup I've seen, the total speed would drop to 100s of kb/s when there was any fog.


its better to have it 50% of the time than 0% of the time... radio waves dont do well in fog either... how would you run ethernet at a distance greater than 500 feet?


Radio is much less susceptible to fog than light because of the lower frequency.

Usually, at any large distances, you'd try to use fibre optic cables if at all possible.


Unfortutnately i post too fast, and dont care about karma... I cant say ive ever downvoted someone over meaningfuly criticical review... ...lower frequency means lower bandwith...wifi frequencies have trouble bouncing around in fog penetrating leaves or even a couple thickness of drywall this is why submarines use ELF... where will you run your fibre optic? on the utility poles, or buried in the ground? how do you connect to someone more than 500 feet from you using ethernet cable how do you mitigate vandalism or theft of your fibre? if you have no connection at all that is inferior to a high bandwith long haul between supernodes that cache each other and are uni directional communicating thus minimizing evesdropping concerns...


The best way to do it is probably laser+radio with additional fiber on high-traffic routes. That way you normally have the bandwidth of laser and radio combined, with slower "just radio" as a backup when there's bad weather, and the fiber bandwidth for backbone connections.


sorry about the delay, im still posting too fast... ...we dont exactly have permission to run our own hauls on corporate/municipal utility poles or trench glass across private property or public R.OW.s radio spectrum is regulated and off the shelf limited so as to minimize the common users ability to step on licensed spectrum...so we have to go geurilla on the long haul...LOS links can be pointed to a multitude of Tx/Rx stations not just one...this is advantageous when the current environment is legislatively hostile...




Applications are open for YC Summer 2019

Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact

Search: