I would caution against just jumping into this and trying to be a real ISP if your experience is in software development, sysadmin stuff, or something kind of tech industry adjacent but you've never worked for an ISP in a NOC environment, as a junior network engineer or similar for somebody else's medium to large sized ISP. There's a lot of missteps that can be made without experience.
In general I would recommend thinking of starting a small ISP as a depth and breadth of knowledge base you need to develop, similar to how an apprentice electrician is expected to work under some more senior/experienced persons in their field for a number of years before being able to operate independently.
I do not want to discourage anyone but I've seen a myriad of 'oh my god why did they do that' small WISPs with anywhere from 50 to 1500 customers. Going down the wrong network architecture/topology/business plan path can be very costly later on if you're 100% learning as you go.
Overall I think a site like this is a great idea, but the sheer scope of things that would need to be written down in a wiki-like format for everything to start your own ISP, and what systems/subsystems/technologies you will want to be familiar with, could fill thousands of pages if printed in book form. It's a gargantuan task to make a website that has everything needed to start an ISP from scratch, assuming the reader hasn't been involved in somebody else's ISP operations first.
Author here. I see your posts often and have a lot of respect for you as someone who obviously has a deeper knowledge of networking than I probably every will. That being said I don't really agree that a person necessarily needs a ton a networking or even any technical background to start a WISP, any more than they necessarily need a ton of experience as a cook to start a restaurant. They definitely need employees or advisors who have that experience, but like you allude to here probably no one can have all of the knowledge to so it themselves upfront.
Also I've also seen a lot of WISPs with a myriad of problems as they grow from a few dozen to a few hundred customers, many of them are able to make that transition, fix their problems, and either grow or be acquired. This seems basically the same as any industry - if everyone waited until they knew exactly what they're doing no one would start anything.
All of this said, starting a WISP isn't for everyone. I talk to a lot of people who are in the early stages of starting an ISP, and often my best, honest advice is that it's not a great idea in their situation.
If somebody has the opportunity to work in a role that has 'network engineer' in the title for somebody else's mid sized ISP first, I would encourage them to think of that like free paid training for a couple of years before trying it on their own.
Of course totally dependent on their level of pre-existing knowledge, level of risk with the capital they intend to spend on a startup wisp, financial partners, and what scale/scope of service area they want to do it in. Varies widely with location. And what competition (if any, other than hughesnet/viasat) exists.
I've seen people do some pretty cool non-profit community based WISP things for under 50 houses, if built right it can be quite trouble free.
The tools and equipment to do really basic 100% fiber based GPON network (or active ethernet) are also a lot less costly than they used to be... The things some guys are doing on really shoestring budgets with FTTH aerial drop cables across roofs and low cost GPON things in cities like Dhaka, Bangladesh show that it doesn't have to be super expensive. In a USA/Canada context, right of way and poles and routes are an issue (even after you're got your ROW, go price how much bucket truck/aerial fiber contractors cost, if you don't do it all in house)...
I read somewhere that 40% of restaurants started by people with no previous industry experience go belly up within 5 years? Something like that. I imagine one could easily spend $400,000 setting up a restaurant and working 14 hour days and ultimately not be successful, so doing something like a startup WISP on one's own similar amount of funds (and loans, private party investors) I would think of something similar to that. If one has the funds and the appetite for the risk and is really determined to do it, then go for it!
> I read somewhere that 40% of restaurants started by people with no previous industry experience go belly up within 5 years? Something like that. I imagine one could easily spend $400,000 setting up a restaurant and working 14 hour days and ultimately not be successful
Absolutely agree with this and everything you said here. I think the only thing I see differently is that while someone should have some relevant experience before starting an ISP (or any business) is that it doesn't necessarily have to be networking. For example, some of the biggest challenges I've seen WISPs face as they try to grow are: obtaining lease agreements for wireless relay sites, hiring and training technicians in large numbers, and in logistics around keeping stuff in stock that they need to grow. Someone with experience from a different industry in some of those areas might do just as well as someone with networking experience.
EDIT TO ADD: One of things I find kind of difficult to communicate with people who are interested in it is that it's just another business, and doesn't really have any better or worse odds of succeeding beyond a few years and making money than any other business, including restaurants. People sometimes seem surprised that you can make money at this, but, of course you can! You can also make money running a taco truck or building custom furniture. But you can also lose a lot of money and time.
I also totally agree that a lot of the challenges are not exactly 'network engineer' related, trying to run a small WISP has so many other things that aren't what you would find in CCNA/CCIE study materials or similar. One can feel much more confident in what they're doing once they have a real firm grasp of BGP, OSPF, various types of MPLS, metro ethernet stuff, optical networking, ptp microwave, linux sysadmin stuff for operational support systems and so on, but that's only one piece of the puzzle of many business process related things to cover.
This is not imperative to do in-house. My company (*shameless plug for Astute Internet) has expertise in those areas you describe, and will provide and/or support the edge network infrastructure for small ISP's in exchange for selling them some of those services (we're a mini-carrier and value added reseller) and charging a nominal retainer to be an available path of escalation 24/7. Due to our purchasing volume and industry knowledge, we are able to sell these services for the same or less than what our customers would be able to negotiate on their own, while getting a lot of support from us in the process.
Although this type of knowledge and expertise is rather specialized, we are able to focus on these specific areas without getting involved in the intricacies of the downstream side of our clients' networks, thus avoiding the need for us to have too much client specific or institutional knowledge. This has worked out very well with the handful of eyeball networks we're currently working with in the Pacific Northwest, and we intend to more actively extend this line of business down the coast in the near future.
A big difference is that you’re selling a service only and a service that is really difficult to differentiate yourself in. A big player could swoop in and obliterate you in a second.
How do you mean? Outside most major cities, ISPs tend to monopolize the playing field and there is great demand for cheaper or more personalized service. Having good customer relations would be a major differentiator, and it's not that hard.
Unless you meant obliteration by lobbying, lawsuits or something similar. There's little hope if your competition goes that route.
You can’t compete with large ISPs on price. They can easily undercut your price for a few years until you’re gone. And a lot of people don’t really chose their provider for ‘good customer service’, they chose fast and cheap.
If you’ve set up in an area that a big ISP suddenly sees as profit potential you’re toast.
Yes and no. Big players will react and offer special deals to people in an area to keep them and lock them in for another year or more if they perceive your ISP as a threat, but if you offer reliable service and good speed and good support, you can eat their lunch.
The problem is that to be competitive you might be looking at years to break even on investment depending on what you offer and where, and how you deliver it. Running fiber on poles to deliver 1gig or 10gig? The buildout is expensive, you have to target dense areas or buildings with lots of units, and even then it's a lot of up-front money and it will be a while before you're out of the red even with a good takeup rate.
> I read somewhere that 40% of restaurants started by people with no previous industry experience go belly up within 5 years?
Experience doesn't guarantee success either. I had a neighbor who was running one of the most successful restaurants at the Mall of America who decided he wanted to do his own thing. He bought a local restaurant and completely remodeled it, and it was great. But it never did the kind of business he expected and he had to close it after 6 months and declare bankruptcy. He told me that he put his house up as collateral for a business loan, and the bank told him he'd need to come up with $700K just to be able to continue making mortgage payments - he was forced to move too.
Sounds like his mistake was likely the costly remodel rolled into the business loan that he couldn’t afford.
For those out there wanting to start restaurants: start small! It’s a heck of a lot easier to go from 10-20 tables to buying the lease out of the person next door and expanding.
If you’re small and full because your busy, then you’ll also generate buzz for being “exclusive” and when you expand, people will still flock there, even though there’s plenty of room.
Source: worked at restaurant that did exactly this. Pizza and decent Italian food. Place is hopping all the time and considered the go-to date night and anniversary spot because it’s hip and trendy. They’ve been open for almost 10 years now.
Oh the place definitely needed remodeling. It had been a restaurant for many years and didn't see an ounce of love in that time. But I'm also certain he went overboard and bought the best of everything. Strangely enough it seemed twice as big, even though he didn't add any square feet.
The location isn't cursed, it was taken over by another restaurant that has been open a few years now.
You wrote: <<The things some guys are doing on really shoestring budgets with FTTH aerial drop cables across roofs and low cost GPON things in cities like Dhaka, Bangladesh show that it doesn't have to be super expensive.>>
This sounds very, very specific. Do you have a link you can share about it?
The WISPs that are started by people with no/little background in NOC operations leave themselves to build, not only insecure, but inefficient networks rather quickly. I started my career on the network side and have worked for a number of ISPs (small and large). And while a WISP creator can make it - they can also just as easily create a bad product unknowingly.
Case in point...
I have family that only had one option for Internet service. Long story short a utility owned a WISP and no longer wanted to run it (they ran it OK, but made a number of mistakes when choosing architecture and products to go to market with). One of their employees bought it out and started making worse choices. The owner clearly had never scaled a network nor did this person have any formal training in networks or the protocols he'd have to manage and monitor.
I've submitted no less than a dozen vulnerabilities within their network and have had to install a monitoring stack for said family members to show the owner of the WISP that during peak hours they were dropping a significant percentage of packets, latency was spiking and SLAs were not being remotely met. The owner has refused to acknowledge any and all data I've provided to them, saying that "things are working well". Yet, I have now a years worth of logged data in a very consumable format that shows otherwise.
The reality is I ended up randomly meeting someone who had worked for them and the conversation was such that the owner was furiously trying to figure out how to fix the problems, but had neither the technical skills to do so or the budget to revamp the architectural flaws.
Starting a WISP almost certainly is not for everyone. Getting it working correctly is only one aspect, but keeping an open network secure for your users is not trivial. Not only do you need a working knowledge of multi-user networks but also a decent perspective with regard to the threat models involved. If you can't, minimally, check those two boxes it seems (to me) a bit egregious to expect others to fork over money and trust you for said services.
Does your site have a section that covers the legal steps a person should follow to protect themselves and their small business? i.e. incorporating their business, business insurance specific to a WISP, processes to follow to ensure legal compliance, what data to log and not log, log retention policies, consumer disclosure, frameworks to put in place for law enforcement requests, agreements to have in place for the upstream fiber provider, etc...
There's a little bit of that in there. Surprisingly, in the U.S. at least, there's not much of a gap between being a legal business entity (LLC, corp) and being able to be an ISP. You don't have to log much of anything (unless you're doing voice services as well). You do have to be able to intercept customer traffic without going on site physically in order to comply with law enforcement requests.
But as someone who is not a lawyer and has no business giving legal advice the best I can do is suggest that you talk to a real lawyer.
I think it might be useful to have a section that includes anecdotal experiences of existing WISP's of what actions they had to take in their location to cover themselves, what legal challenges they ran into and any caveats or scenarios to be aware of. Obviously with the disclaimer that it is not legal advise and they should retain their own legal council. It might give folks a better picture of what to expect before they consider the investment.
For what its worth, I really like the idea of people implementing some options for those that might only have access to one ISP. I was locked into Comcast for the longest time and would have loved to see more options available.
Legal requirements can be funky. Back in ~2007 or so I considered doing my own WISP just to get any reliable internet that wasn't making 300 baud modem look nice (put enough latency on telnet and it will look like that).
Turned out that if we wanted to cross a community (smallest administrative region in Poland), to be fully legal we might have had to submit war and disaster preparation plans to Ministry of Defense. For cooperation with military and other services, just in case.
I am not an expert but I've seen this book recommended: Small Time Operator: How to Start Your Own Business, Keep Your Books, Pay Your Taxes, and Stay Out of Trouble
What about the customers though? My family used to be chained to a WISP, and we paid $100/month for internet that made dial-up look like a viable competitor. Not to mention, downtime was atrocious (1/10th of the time the service would just be out).
I think the bigger argument for not starting a WISP is that nobody wants to be a guinea pig in your amateur networking project.
(Author here) WISPs are more like local restaurants than big chains. Some local restaurants are terrible. At least with a big chain you know what you're going to get.
That being said, the WISP you had probably just didn't know what they're doing. Which is true of a lot of WISPs (and small restaurant owners!) And is actually in large part what motivated me to write this guide! I'd like to believe that if they followed the guide, and/or gave me a call to help out with the parts they're struggling with, we could fix it. I'm personally involved in running several WISPs right now, and without exception we are providing 100mbps+ service, <20ms latency, and at least 99.9% uptime measured to the customer.
FWIW, and not to undersell the challenges, but I think you're extrapolating too much from a single example (and possibly an old one). The started-tiny-and-local WISP around here is $100 a month for 50/50 and it's been pretty rock solid. While I certainly haven't run a WISP myself, I have set up point to multipoint wireless networks for some local businesses, and with current equipment it's impressive how straight forward and relatively turnkey the basics of the physical link itself have become. It was no problem to have dozens of points over distances of 5-16 miles with solid 100-400 Mbps links. My understanding from far more senior people and reading about real WISPs is that there is a non-linear increase in difficulty/expense over greater distances, scaling up to hundreds/thousands of points, and handling things for multiple independent entities rather then what is effectively a big LAN even if sectioned up with VLANs. There's also completely non-technical challenges to be ready to deal with like compliance departments and legal/law enforcement interaction aspects of when a customer inevitably does something naughty. The path has been tread but it wasn't anything I've had to deal with.
Still, if I was still in an area with dial up speeds and Starlink wasn't a thing, helping jury rig up neighbors with a mini-WISP is a project I'd certainly consider contributing to and believe could be quite reliable and easily beat out DSL at least.
Also speaking of Starlink, I'll be curious if a side aspect of them would be driving phased arrays in general down in price significantly. That'd be a pretty interesting development for terrestrial PtP/PtMP imho, since a lot of the trickiness over long distances is precise aiming. If the antenna merely needed to be aimed with 10-20° and then could perfectly align with the other from there, including real time adjustments for some motion or interference, that'd be another pretty cool improvement. And for the time being at least Starlink has left an opening in that the connections are quite asymmetrical (ie., 200 down but only 30-40 up), which is challenging to work around given their physical and economic constraints. Whereas a WISP can be full symmetric right up to gigabit speeds, assuming their backbone supports it. That could let some remain competitive beyond just density limits.
The phased array wisp thing has been tried before and failed. People seem to think that wisp is somehow using different physics than 4/5G. In reality it's more like Joe's taxi vs Uber.
>The phased array wisp thing has been tried before and failed.
Examples? I don't know of any phased array PtP/PtMP system, at all, in anything like the form factor and pricing of Starlink or something like Ubiquiti's airMAX/airFiber/LTU series operating in the 5, 11, or 24GHz class spectrums. In fact your whole comment makes zero freaking sense, what do you mean "phased array wisp thing"? I don't see how it'd fundamentally be any different then all the very successful WISPs I'm familiar with, or for that matter the LR-LAN stuff that I've done myself. The advantage would be in that one of the trickier bits at longer ranges is getting alignment just right, labor or kit correcting for shifts, and tradeoffs in allocating regulatory max transmission power budget across end points. Skilled personnel can of course learn the alignment part reasonably and an important differentiator for a long time amongst different kit was the kind of tools it had to help with alignment. But it's not nothing and it can take a while, and then there is the rest.
A phased array system like Starlink's that could make its way into $150-500 CPE and hopefully <$5k basestation would be a nice evolution. Only need to get it roughly aimed in the right direction and then the system does the rest with no further human intervention, zero skill required. Basestation could extremely rapidly reallocate high power to various end points which matches consumer load patterns well rather then having to pick 30°/60/90/120/3x30/whatever sector antennas or 4'+ dedicated big boys for real distance.
I have no idea where this "4/5g physics" thing came into your head from. We're not talking physics we're talking economics and barrier to entry.
Is Starlink such a big deal that it would materially affect phased array costs? Cell base stations are already ubiquitous. From my own experience with phased array systems, the big cost is usually in the massive FPGAs that are needed to process all the incoming data in parallel.
>Is Starlink such a big deal that it would materially affect phased array costs?
I guess I think the Starlink terminal is technically one of the coolest and least appreciated aspect of the whole system. Sure it's not a military AESA system but getting that level of phased array into a 100W all weather CPE with a $500 target price they plan to produce tens of millions of is something I don't think has ever been done before, and it definitely could blaze a trail in terms of economies of scale there. SpaceX's rockets and sats seem to get most of the press, and are certainly critical, but the terminal is also a big deal IMO.
I mean, think about it: they've built an end point that will do a point to point link of 500+km to base stations with relative velocities of kilometers per second and can user-imperceptibly swap between multiple ones in real time while maintaining a bandwidth of hundreds of megabits per second. For $500 (granted it's probably costing them at least double/triple that right now, but they aren't making tens of millions yet either). A much more cut down, cheaper version of that could still make a number of terrestrial PtP/PtMP links a great deal easier for random people to take a swing at and get even better results with then now.
Again, I'm certainly no expert at all. But I've kicked the tires on a little of this stuff a bit, recently got my first chance to work with second gen 60 GHz links for example and those feel like a certain amount of work to get just right over multi-km ranges. But if for another $100 suddenly they just handle it themselves as long as they're within 10-20° horizontal/vertical? It'd make some aspects of installs more accessible and cheaper which doesn't seem like a bad thing to look forward to?
OP is of course right that none of this addresses the many other aspects of doing networking.
Pivoting your server or software stack is like changing a floor in a building.
Changing your network stack is like changing the foundations of the building.
Sure you can do it, but the bigger your building, the harder and more expensive it's going to be.
There's a lot of value in learning a lot of these basic mistakes first on someone else's dime.
Sure, you don't need to be a trained chef to start a restaurant, but if you dont know that undercooking pork is going to make your customers sick, you probably should be hiring someone else to do the cooking.
I think your analogy with restaurants is flawed. There's a reason a lot of newly started restaurants fail.
You most definitely need some form of experience from the restaurant business to succeed as a restaurant owner, unless you are really lucky. Your analogy make it seem as being a chef is the only job that exist in the restaurant business.
Sure, you can succeed without experience, but your chances to do so increases dramatically with experience.
I realized after the fact that I didn't elaborate on my meaning here, but it seems like you picked up on it.
What I meant was: You should have experience with at least one and preferably several of the important parts of running an ISP before you start one. NOC experience is one of the things that you could already have experience with, but isn't the only one.
> Your analogy make it seem as being a chef is the only job that exist in the restaurant business.
This actually makes my point better than I did. Being a [chef/network technician] isn't the only job that exists in a [restaraunt/wisp].
What type of profit potential is there to doing this? Seems like good stable income, but I also see lots of downsides with literally not knowing how anything works (and not being an engineer). There would be an awful lot of new information to learn, so the profit potential would really need to be there.
That would vary wildly by geographical location and whether it's some place that has no WISP at all right now, population density, competition, terrain and tree cover, number of houses, etc. I'm skeptical about the idea of starting a very small WISP right now, one could easily spend $40-50,000 on basic infrastructure for a small area and not be able to compete in nines of reliability/uptime and speeds with Starlink.
Exactly, you need to have access to tall buildings / towers etc with fiber. The cost of putting a couple of antennas on a tower can basically eat up any profit you would get. You are also competing with LTE/5G, and one storm can wipe out half your infrastructure in an afternoon.
Heh, I can't imagine that feeling when you have really no idea what you're doing, you have 50 paying customers, and then all your equipment goes down. This sounds nightmarish unless you are some sort of electrical engineer
As a one-man WISP responsible for everything at the physical level and the logical/IP configuration of the network, monitoring systems, billing systems, customer support systems etc you are also trapped in a personal hell of never being able to go on vacation and on-call 24x7x365.
There are certain sorts of people who are capable and comfortable with doing that and have the motivation to do so, but running a small sized ISP realistically takes 3 to 4 people with differing but complementary skill sets in order for everyone to not burn out. It takes a fair bit of revenue just to handle what would be the reasonable payroll cost for that plus operating costs plus overhead.
My first company was an ISP. There were 4 of us. We were around 19. No business experience. No networking experience. How hard could it be, right?
I configured my first router by calling our provider and saying we'd set things up but something was wrong, and having them echo to me what they did on their side so I could "check" our side - they charged for the setup, but this way we got it for free.
We literally lived in the office for the first year, and one of my most entertaining experiences was answering the phone at 2am on a Sunday morning and the person on the other end expressing surprise that someone answered - they'd just called out of sheer desperation and expected it to ring forever.
Then there was the person who spent three hours terrorising us before it transpired he'd not connected his modem, nor turned it on, or in fact taken it out of its package - something the person talking to him found out call by call, while the rest of us tried to not audibly laugh, because surely nobody could be that stupid? (a "learning experience" - it was how I came to understand why support people ask all those really obvious "stupid" questions.
I can't make up my mind if all the network issues were the worst part, or if the support calls were.
This was 26 years ago, and I have just about gotten over the soul-sucking parts of it by now.
I even contemplated setting up a local ISP again... I don't know why.
> Then there was the person who spent three hours terrorizing us before it transpired he'd not connected his modem, nor turned it on, or in fact taken it out of its package
I worked at a local mom and pop ISP at around the same time, probably a few years later. Those sorts of customer calls were the best (after they were resolved of course.)
* One guy would call at 2 or 3 AM and scream into our voicemail "Internet's DOWN! FIX IT NOW!!!!!". His wife would call the next day and apologize.
* One time, someone called irate that their dialup service wasn't working. He called the number we gave him, and all he heard "was a bunch of ringing and static". After several minutes we figured out he was dialing the modem bank from a telephone and expecting to use the Internet without a computer. He did not own a computer, and was quite upset we had not told him he needed one...
* It took two or three hours one time to figure out why someone couldn't connect. Helpfully, they figured it out themselves -- they were entering "the letter zero, not the number zero".
* A guy called up one time wanting to tell us about an interstellar propulsion device he developed (which also, of course, could cure cancer if the 'magnetic field' was reversed.) I suggested he call the Jet Propulsion Labs and gave him the number.
For me, the support calls were always the worst and I was very glad when the company got big enough that I no longer had to regularly take calls (although the ones I did get were the worst of the worst.)
I couldn’t not refer the guy after I spent 15 minutes listening to his…not exactly well-thought-out, but certainly extensively-thought-out idea. Also, it was Friday afternoon and I needed a way to get off the phone. Just hanging up seemed rude.
I once lost internet for a week when our WISP provider's base station went down. We tried calling his company for days, and he only came to fix it when we found his cellphone number and accosted him personally. Thankfully we use Starlink now.
I thought this was going to be a game reliving my experience as the System Administrator at our local dialup internet service back in the 1990s. I wasn't there at the start, but I ran the thing for 2 years, serving about 200 customers. I wrote the "Internet Installation" floppy disks that people used to install networking on Windows 3.1, and configured dialup icons on their desktop. I was the ONLY technical support they had.
I learned that people don't use the same words we technical folks do, so if you get a support call and ask "Have you installed anything?" the answer they give is no, but, if you ask "do you have any new programs or stuff", the answer will be yes.
Why? - My theory is that they either didn't know what "installing" was (nobody opened up the computer to put something in), or they themselves didn't do it.
You have to have a very strong ability to route around the person on the other end of the phones preconceptions of the world and get the right answers anyway.
If you're going to be an ISP, you're going to be on call 24/7/365, and you're going to see weird shit. More than once we had to deal with people wanting to post adult content for money.... and this was back in the days of dialup. Who knows what the heck the folks are doing today.
> You have to have a very strong ability to route around the person on the other end of the phones preconceptions of the world and get the right answers anyway.
Back in the day, I would get called in to support a SCSI-connected external device. If you asked "Is the cable connected?" the answer was always "Yes", and sometimes "Do you think I'm a moron? Of course it's connected!". So we started asking them reach around the back of the device and unplug the cable, wait 10s, then plug it in again. This was usually followed by a sound of surprise as they realized that, wotta ya know, the cable wasn't connected.
A cold-boot is a different request from a reboot. There are devices, like a crappy old DSL modem I used years ago, which do not fully re-initialize state on a software requested reboot.
(Author here) This is great, you can't see me but I'm laughing with you! :)
I have sometimes considered making this in to a game, actually. It could have some fun territory and financial mechanics. But the more I think about it the more it seems like work, and the less I want to play it!
To be realistic it would have to wake you up at 3AM and make you pull pants on and drive out to the bottom of a tower in a muddy field that smells like chicken shit to drop in a generator. While it's <0 F˚ outside.
Hi folks, author of startyourownisp.com here. Not sure why we hit the front page today, but happy to be here! I'll try to hang around and answer questions for the next few hours. Thanks!
Thanks for the article! Well written I think. I agree with the top post that if you're a engineer who hasn't dealt with low level networking stuff, worked in a DC, or had to be on call, this probably isn't a good first project. But golly, I can't wait to have the free time and capital to try starting one. (I live in a spectrum dominated area) Maybe try and make the whole operation solar powered.
Also fun, for anyone else who's interested taking "DIY self hosting" to new extreme, it's possible to start your own cell phone company! I don't have a link handy, but search "starting an MVNO" will yield tons of links. When I did the math, it worked out to require only ~100 customers to hit break even. (assuming your time and labor is free, of course)
I have unreasonably happy giggle-fit inducing day dreams about some day using my own designed and manufactured smart phone (easy, because IDGAF about trendy features. running stock debian is fine.), which gets service from my own service provider, and my homemade laptop, getting internet from my own ISP, in my solar powered house.
Also also, for anyone who lives in the SFBA, monkeybrains internet is the best ISP evar. I was one of their first customers, and I miss my ~4ms ping times to the office, and ~8ms ping time to my colo box.
Thanks! I know some of the folks at Monkey Brains, they're a great ISP and super cool people too!
> I agree with the top post that if you're a engineer who hasn't dealt with low level networking stuff, worked in a DC, or had to be on call, this probably isn't a good first project.
I don't really disagree, but I would just say that if you're interested in doing this, and have money you'd like to invest in it, reach out to me. With our powers combined we can probably make something work!
Every time I’ve seen your site posted here or on reddit you’re always in the comments offering to help and answering questions so I just wanted to say major kudos to you for being so helpful and responsive. I’m sure it can’t be easy but you’re doing a great public service by sharing your knowledge.
Thanks, appreciate the kind words. It did derail my plans a bit to hit HN this morning, but it's fun to chat about and HN has done a lot for me over the years!
Related in both cases. SYOISP describes what I, as a somewhat opinionated author, believe to be the absolute minimum amount of information required to build a business that can be called an ISP. To the extent that the companies involved in meshcenter.org and also Starry are ISPs, there is definitely overlap.
> SYOISP describes what I, as a somewhat opinionated author, believe to be the absolute minimum amount of information required to build a business that can be called an ISP.
Thanks. Then, what'd you say are key differences between meshcenter.org and syoisp? Curious because meshcenter goes for maintainability and affordability too, from what I gather.
Also, has Facebook's foray into the space with https://telecominfraproject.com (and Google's too with https://opennetworking.org/) change the landscape? They're going after 4G/5G like deployments, which is in sharp contrast with what starry and meshcenter are doing?
> Then, what'd you say are key differences between meshcenter.org and syoisp?
No big differences, everything is applicable to both. In practice 'Mesh' has kind of become associated with 'non-profit, volunteer staff,' and SYOISP makes the implicit assumption of a for-profit entity.
> Also, has Facebook's foray into the space with https://telecominfraproject.com (and Google's too with https://opennetworking.org/) change the landscape? They're going after 4G/5G like deployments, which is in sharp contrast with what starry and meshcenter are doing?
Unsure yet what these efforts (and I'll add Microsoft's Airband) will amount too. I will say that I don't really see them tackling the actual hard parts of improving connectivity, which is primarily the actual building of physical infrastructure. I mean, other than Google obviously doing so with Google Fiber. Software can only get you so far in this effort, which is I think a lot of the reason that the incumbent providers have maintained such a stronghold over modern tech companies.
I'm working on connectivity stuff at Facebook. Facebook is doing quite a bit more than just 4g/5g like deployments but the focus is primarily outside the US. We do have some public tools that I think are especially relevant to your effort: https://www.facebook.com/isptoolbox it's actively being worked on and there are many unreleased features in development but would be happy for any feedback too. Hope it's helpful.
Probably not:
>We designed every piece of hardware ourselves to ensure it works seamlessly to beam internet from the tops of towers to building receivers and right into your home without a hitch.
(author here) It's similar in architecture, if not in the actual hardware being deployed. FWIW my personal opinion (informed by personal experience) is that designing your own equipment is a huge distraction with little benefit. We'll see how it works out for Starry.
As I understand Starry has moved to focus largely on apartments buildings and in-home WiFi (rather than residential Internet service to single-family homes.)
Is this a reasonable way to make some money? Or for people who are fed up with their ISP options and are ready to take matters into their own hands? Basically, who do you imagine the typical user of your guide is?
> Or for people who are fed up with their ISP options and are ready to take matters into their own hands? Basically, who do you imagine the typical user of your guide is?
If your only motivation is that you are fed up with your ISP (join the club!) that may not be enough. If you are fed up with your ISP, have a spare $25k laying around, and/or have access to grant money, then we should talk. (Many states have grant funding available today specifically for broadband projects in rural or underserved areas.)
I do consulting work on these types of projects as my main gig now. Many of my customers own businesses in other industries, live in a small town, and decided that their Internet sucked and they want to fix it. But it does take some capital.
Having done this informally many years ago for a dozen neighbors in a village before we got decent broadband I can relate. People get real weird and twitching when there internet is out and there is nothing like being woken up at 6am on a Sunday with a "customer" in you bedroom ranting that the INTERNET IS DOWN. Or having to bust the kid who hid a wifi repeater in the village library to torrent gigabytes of "ISOs" over our tiny 1.5Mbit connection
I moved to a new apartment and forgot to set up an account for electricity so my power was shut off for a day and a half. The most interesting thing was how it actually didn't bother me too much because I still had my phone with mobile data, if I didn't have any form of internet connection that would have been _much_ more of a hassle since I wouldn't have even been able to access the energy company website to resolve the issue.
My family bought internet from a small ISP nearby who had done the same thing. We paid ~$100/month for WISP access, with ~10% downtime, 500ms of latency (at lowest), and 500kb down. It was a nightmare, and my siblings performance in school started tanking as they couldn't access homework or lectures from home. We eventually switched to Starlink, which has been a 100x improvement, both literally and figuratively.
Please, nobody ever start your own ISP. It absolutely sucks to use your infrastructure.
Author of startyourownisp.com here. Your experience is unfortunate and definitely not unique, but also not universal or I would say even the norm for WISPs. I'm personally involved in running a few different ISPs in rural parts of the U.S. and Mexico. In all cases we are providing 100mbps+ service with <20ms latency to most of the Internet.
I'll take your word for it, but I'm definitely not switching back from Starlink any time soon. We had more issues with our old provider in a week than we've had in 3 months of Starlink service.
It's a small miracle our local isp stayed afloat. Service was so flaky for a decade and a half that I could tell the weather outside by the lagspikes and speed drops. A few years ago he won some national/EU grant and now he operates a village wide fiber network.
Probably the biggest infrastructure upgrade since the introduction of tapwater/sewage in the 70's.
Related and fun: the Fremont Cabal Internet Exchange[1]. The person who started it has a great blog[2] about all kinds of related things. There’s a really good interview[3] with him about how it started and what it takes to run it on the On The Metal podcast (highly recommended in general)
Over the last year or so the company I run moved some content delivery workloads from a CDN onto our own network. So now we have something like 100Gbit/s of total external bandwidth.
The next thought in my addled brain is that since we only really use the egress side of it, perhaps some of our neighbors would enjoy affordable 10G internet access…
(Author here) Maybe possible to figure out a way to use this. The problem is the people using it would still need transport to your network, which in practice can cost as much or nearly as much as direct DIA to their location.
The reverse here is maybe interesting too - a lot of WISPs have a bunch of spare 'upload' (from their perspective) bandwidth available, and by definition it's very geographically distributed, maybe that would be of use to CDNs?
I'd love to chat more about this if you're interested. Email in my profile.
The tricky bit is that a CDN doesn't really want geographic distribution so much as to be close to network distribution points. If a WISP's customers aren't enough to warrant a CDN node and their internet backhaul is fiber to a not particularly nearby internet exchange point, there's not a whole lot of benefit to the CDN to be at the WISP rather than the internet exchange.
It might be different if the WISP has connections to local residential ISPs that aren't well connected to local internet exchanges (or there nearest internet exchange isn't very near) so the WISP facility offers a way to get (network) closer to more users.
I could also see some potential for mixing traffic streams between a WISP (or several) and a CDN to try to balance traffic flows enough to qualify for peering with networks with restrictive peering policies; however, they also often have geographic requirements that might be harder to meet.
Thanks, this helps to confirm what I've always expected from my end (the wisp side.) It sure seems like a good match at a high level but seems to fall apart once we get in to the details of making it work.
> You probably need less than you think. A 1Gbps fiber connection will easily serve 500-800
No thanks. I was intrigued, but ultimately I don't think my community is better served by having yet another oversubscribed Comcast-like ISP ( even though its run by individuals )
Over subscription makes complete sense because virtually no one maxes out their connection all the time so it makes sense to not reserve the maximum speed for every customer.
My ISP lets you see a chart showing the nearest node's current usage and its capacity so you can see it just tap the top of the capacity each day at peak streaming hour and then spend most of the day almost unused.
All IP networks are oversubscribed, even the ones charging "datacenter transit rates". The difference, in this respect at least, is that good providers monitor utilization and proactively upgrade capacity.
It would be interesting for ISPs (especially municipal or small ones) to have a live graph showing ingress/outgress on their various routers and lines.
There’s also “soft” limits (currently we pay for 10G on this fiber line that could do 1000G) vs hard limits (we are at 1G on this 1G copper connection).
I have 320 subscribers (on fibre) out of one of our PoPs, and the evening peak is around 1.6-1.8 Gbps these days. Oversubscribing 800 customers on a single 1 Gbps connection is going to result in an absolutely awful end user experience.
Most WISPs make the mistake of failing to monitor airtime on their access points. With much of the low cost gear on the market the performance of a sector is going to be limited by the performance of the customer with the worst SNR. Newer gear that implements MU-MIMO significantly improves that since the access point can transmit to multiple subscribers at the same time. Have a look at some of the Cambium equipment. Other vendors have taken to implementing better QoS on the wireless phy side of things, but without MU-MIMO, it's just shuffling chairs on the deck of the titanic while the whole network sinks to the level of the guy with a bunch of trees in the fresnel zone of their 5 GHZ antenna.
Also just a note for those curious: this is how I like to measure what's needed for a WISP, and track how it changes over time.
Set up a 1gbps connection and start adding customers. Watch the usage on the circuit as your customer base grows. You want to make sure that any time a customer comes online, they can get their full speed according to the package you're selling. So if you're selling 100mbps packages, you always want to have at least 100mbps available on that circuit, even at peak times of day. As you add customers, you can start to estimate how much traffic, on average, each customer adds during the peak time of day (it's not as much as you'd expect.) Then you can estimate how many customers you'll have before you approach your limit (1gbps minus max speed package.)
You have to monitor peak utilization of the network on the order of seconds to minutes. The rule of thumb from queuing theory is that network performance tends to tank when capacity utilization hits about 80% => a 1 Gbps link is only good if your max utilization is 800 Mbps. But if your monitoring only looks at 1 hour averages over the course of a month, you're going to miss the periods of seconds where everyone's getting ping times of >100ms and having their VoIP calls drop out.
There are a lot of things you'll learn along the way if you try to run an ISP. =-)
commercial isp's do oversubscribe in the range of 1k-20k. it'll work fine if there is some form of fair-queueing that limits interference between customers.
It seems like there should be a lot of potential for something like this in cities. The site says that line of sight and apartment buildings are problems but in the east coast cities I'm familiar with there's lots of neighborhoods consisting mainly of row homes or 2-3 story multifamily houses. To me places like neighborhoods like those would be a natural fit, more so than suburbs.
Also for me the main appeal of this is the potential for a not for profit, co-op style ISP. I.e. owned and operated by the customers. Although if the fiber is still coming from a Comcast or Verizon I'm not sure how different it would be vs. just buying consumer internet direct from Comcast/Verizon.
Could be! East coast has traditionally been hard for WISPs because of trees. Recently I've been doing a proof of concept in Georgia, though, and had a lot of success even with tree cover. Would love to try out a few more of these places that have traditionally been hard for WISPs. (Author here.)
Had a colleague that ran an a very small scale ISP. It was back in the day when the country is transitioning between dialup and broadband. Needless to say that it was very short lived.
Nice. I started an ISP in 1993. I was not a network engineer. I was an automotive engineer. It was the most intense learning curve I have ever been on. Back then you could find lots of people willing to help out who knew everything about networking. I went on later to dabble in ecommerce by publishing the ISP Business Plan. A word template and spreadsheet to get you started. Dozens of ISPs around the world used it when they started up. This site feels like the extension of that. Still a good business to get into if you don't mind being stressed by the weather. :-)
This is so interesting! I noticed a sign while driving yesterday that was unbranded saying "high speed internet" and an arbitrary local phone number. I presumed it could be a reseller but my mind started turning on how I could create an ISP and what that process looks like. Next morning I see this.
I don't think I would execute on this personally because of the support required -- the spreadsheet takes into account the building but less so the maintaining. I would struggle to be hated as much as people fume at ISPs when their service is impacted.
While this is something I would never take on for a handful of reasons, this is extremely cool in concept and this would be an incredibly fun project to put together.
To be honest, the first thing that crossed my mind was "good luck getting an IPv4 range in 2021". ;-)
The second thought was. Don't bother unless you've got a clear competitive differentiator. There are already too many ISPs out there as it is without another "me too" cluttering up the market.
> isn't the real world not yet ready for pure IPv6 ?
Unfortunatley not. I can name a number of mid-to-large sized ISPs in major "western" countries who are still not able to give their customers IPv6, and the worst offenders haven't even finished deploying IPv6 on their core networks !
Fortunatley that number is diminishing as the months and years go by, but for now we are still in a world where you're going to struggle setting up a new ISP if you don't have any IPv4.
I helped build a couple of ISPs in the 90's, back when upgrading from a 56K leased line to a T1 was considered a right of passage. Fun times. Not sure I'd want to go back. Seems hard to compete with the cable company in high density areas. Maybe a WISP if I was out in the boonies...
Author here. This is definitely something some WISPs are worried about. It's likely that Starlink will take a lot of the most rural customers. WISPs can also be very successful in more suburban environments, and can speeds much higher than Starlink can provide in suburbia (300-500mbps) at a lower cost, so there will probably be a place for WISPs at least until there is no gap between fiber availability and Starlink as the best option.
Also, a lot of WISPs have started running fiber in rural areas, so that's another way they can stay relevant.
starlink is in no way a competitor to a modern ISP, on a modern ISP it's normal to have >=1Gbit and more than 10 times less of a latency compared to starlink, and the maximum potential userbase for starlink is a fraction of internet userbase
oh, my bad, i did open the link, but jumped straight to marketing section(since it was my point of interest)
and somehow assumed it was about local small ISP's, with wireless it's not that simple, yes.
But still, it's mind blowing how horrible ISP's in US are, you would think if in some third world countries it's accessible, everyone in US should have it.
I honestly feel like starting a WISP is risky right now. Companies like Starlink and T-Mobile have the money and infrastructure to compete and solve a lot of problems that WISPs have.
Please remove my memories of an acquintance who lived and slept in a warehouse in Northeast Minneapolis (before that part of town gentrified) running Intuitive Networking Solutions (INGS) during the mid- to late-1990s:
I'm curious, do you happen to have any stories and stuff to tell about him? (Hope you live on the first floor so you can still use the defenestrated laptop)
I work for a wisp as a software engineer. I'm knowledgeable enough about our network/prices. None of the network engineers believe the numbers spacex is claiming they can do yet they can see it right in front of their eyes. The main argument is, "Well once this network actually gets any load it'll crumble." Or, "The latency is going to be through the roof." But people who live where a WISP is their only option already don't care about latency all that much.
I can't definitively say this is going to be disruptive but I can tell they're nervous. From my perspective the future of rural internet is in space.
I work in networking but not for an ISP. I have worked in networking for a long time.
To me, Starlink looks like a very typical Musk venture - vastly, vastly overhyped. As someone who would like to move to a more rural area, it appeals to me, but the numbers don't work.
The _total planned switching capacity_ of the starlink constellation, and I emphasize planned because the current capacity is kind of a joke, is approximately that of a single current generation multi-linecard network switch. And that's assuming they do in-space switching, which at the moment they do not, and to my knowledge none of the current satellites are even capable of doing so in a non-experimental production capacity. They also have very significant routing challenges if they ever do that (and even if they don't). Worse, the skinny end of the pipe is on the wrong side of the satellites, so even things like caches are not really going to help them.
It just isn't even remotely possible that Starlink is going to provide the current level of service once they get any kind of significant user uptake. The bandwidth per spot is too low and the spots are too large.
I personally think that half the Starlink story (inter-satellite switching) is going to be SpaceX's "full self driving." Unlike FSD, SpaceX has an easy out - they will build out enough ground station coverage that they don't need to do inter-satellite switching except for the air/sea cases and those markets they may simply abandon.
The true innovation of Starlink is that their ground stations are dirt cheap, but that's not a sexy story.
Everything in this field is about how much of what capacity of medium you have per subscriber. FTH has 100% of a piece of glass. WISP has ~10% of a 25MHz channel with ~80dB noise floor. Musk has ??% of a ??MHz channel with ~90dB noise floor. Apply Shannon and you have your answers.
https://freedomfi.com/ is selling hardware later this year which will allow you to become your own cell tower for 5G CBRS bands. You'll get paid $.50 per gb consumption, paid in Helium HNT crypto (yes, crypto). Its still a big work in progress but could totally unlock private deployments of 5G networks and make them almost as easy as setting up wifi.
Starlink doesn’t change anything, it’s not for dense coverage. Not too much overlap with WISP. Revolutionary in other ways though.
The main problem with this way of transmitting is that you are on unregulated parts of the radio spectrum. So there’s nothing stopping someone from jamming your equipment, grinding everything to a halt, whether it’s on purpose or by accident. Maybe in practice it’s rare, but it’s too big a vulnerability for me to invest in such a business.
I was wondering about laser modems. It could never be jammed without some kind of expensive and conspicuous physical intervention. I would feel much more comfortable investing in that even if lasers were more expensive or a pain in the ass.
Not that it matters. 5G is going to put all these guys out of business.
> The main problem with this way of transmitting is that you are on unregulated parts of the radio spectrum.
Traditionally this is true and is still mostly true today, but a ton of WISPs picked up PAL licenses in the CBRS band and are using that. Also, as you mentioned, in practice "jamming" equipment is uncommon (and very illegal!)
Also I always recommend that folks use licensed or FCC registered links for their wireless backhaul, which limits the amount problems you'll get from interference.
> I was wondering about laser modems. It could never be jammed without some kind of expensive and conspicuous physical intervention. I would feel much more comfortable investing in that even if lasers were more expensive or a pain in the ass.
I've used some of these device (called "open air optical.") They're ok for very short range connections, but over anything more than a few hundred yards the performance of wireless is much better.
> Not that it matters. 5G is going to put all these guys out of business.
As a WISP you have the option of deploying 5G networks. It's expensive, but might be worth it. Many, many WISPs (myself included), operate fixed 4G LTE networks, and 5G is the next step. I haven't dabbled in 5G yet, but I have sales people calling me somewhat regularly ready to sell the gear. To the extent that 5G will live up to it's promises, WISPs are well positioned to take advantage of it.
Freedomfi (https://freedomfi.com/) is selling hardware later this year which will allow you to become your own cell tower for 5G CBRS bands, based on Facebook's magma project. You can backhaul a 5G modem/antenna, offload to a major MNO/MVNO, and get paid $.50 per gb consumption, paid in Helium HNT crypto (yes, crypto). Its still a big work in progress but could totally unlock private deployments of 5G networks and make them almost as easy as setting up wifi.
I am already working full time remotely in the middle of the woods where there is "no cell service" getting 6 mbps up and down on an unlimited plan with a super high gain antenna aimed perfectly at the tower through several miles of trees. Once 5G modems come down in price my throughput will only increase. Coverage is already amazing on the under 700 MHz LTE bands even with carriers that traditionally have had a reputation for terrible coverage - even when the tower cells are aimed in the wrong direction! (They are aimed at the town and highway nearby, away from me.) High gain antennas are the most important component of my stack.
11 dBi Yagi Antenna for TV White Space (470-862 MHz) is my choice for extreme penetration. It's several feet long.
I don't trust the specs of the parabolic antennas on the market. They are supposed to be better but I have heard from multiple reviewers on multiple products that their specs are wrong and they simply don't deliver high gain at low frequencies when advertised to be tuned for 4G LTE. I think it is because to cover 4G LTE bands fully you need to support a ridiculously wide range of frequencies from 600 to 2200 MHz.
My low frequency LTE yagi is awesome for difficult wooded areas. It can shoot through at least 3 miles of trees.
I think your Proxicast peaks at 11 dB gain and you need to check which frequencies that's valid for. With such a large frequency range spec'd on that you can't be sure you have high gain on all frequencies. It can drop off drastically on important bands.
That's why I have a dedicated yagi for low frequencies and a similar antenna to what you have for ~2 GHz bands.
I found through experience the best proxy for correct antenna aim is an upload speed test. And the best way to determine aim is by using cellmapper.net and calculating a compass bearing from your location, then using a compass (app) to aim your antenna. Not the more high tech metrics reported by your modem like RSSI and link quality. They can be very misleading in general and especially when the link is idle.
> 5G is going to put all these guys out of business
Does 5G make much of a difference here? From what I've read, in the the bands that give you reasonable range (mmWave doesn't work at these distances), 5G at the very most gives you like 20% more bandwidth than 4G LTE. That doesn't seem game-changing.
In most of the places where I know people who use a WISP, they barely have 4G coverage, and when they do, it's not a strong signal to say the least. Most of the places (at least in the US) where 5G is going to compete are probably places that already have cable.
Exactly. 5G doesn't focus on increasing coverage but mostly latency. The mmWave stuff with the 1 Gbps+ speeds has a coverage of around a block, so that will only be used in super-dense areas.
From the competition section of the site, the sweet spot for WISPs seems to be just under suburban, where there's enough density but not quite cable coverage (or the local cable monopoly isn't giving their network enough TLC)
> Exactly. 5G doesn't focus on increasing coverage but mostly latency.
This is my opinion as well - 5G will improve latency and also capacity (meaning carriers won't have to be as stingy with tethering plans) but not necessarily top speeds or coverage area.
mmWave won't penetrate a door or a window or a tree. It really doesn't work in rural areas at all which is where WISPs make sense. And WISPs don't make sense if the local geography is rocks, hills and trees that make it impossible to get a clear line of sight.
I imagine 5G would be even worse since it has less range. The 4G spec supports speeds higher than most Americans are getting on fixed internet services. I'm not sure why everyone treats 5G as the solution to all problems when it's primary use case is getting mobile usable in ultra dense situations like sports stadiums.
> So there’s nothing stopping someone from jamming your equipment
Nearly a decade at a small/medium ISP in Brazil (where illegal competitiveness isn't uncommon) and I've never seen this happening to us (or if it did, we never found out).
This is because jamming the spectrum requires a very powerful antenna that will _also_ interfere with every other antenna in the region, including probably their own. Instead of that, it's way easier to just grab some scissors and cut off your competitor's wires (which has in fact happened to us on numerous occasions).
So while there are anticompetitive practices I don't think this is one you'd ever have to worry about.
Built an ISP back in the mid-90’s with some friends as a startup. Dial-up, ISDN, Frame Relay, DSL, etc.
My one piece of advice is to not overestimate sales. We engineering types tend to underestimate how hard it can be to sell things. We love the “build it and they will come” fantasy. The real world rarely works that way, especially for something as run-of-the-mill these days as internet service.
In general I would recommend thinking of starting a small ISP as a depth and breadth of knowledge base you need to develop, similar to how an apprentice electrician is expected to work under some more senior/experienced persons in their field for a number of years before being able to operate independently.
I do not want to discourage anyone but I've seen a myriad of 'oh my god why did they do that' small WISPs with anywhere from 50 to 1500 customers. Going down the wrong network architecture/topology/business plan path can be very costly later on if you're 100% learning as you go.
Overall I think a site like this is a great idea, but the sheer scope of things that would need to be written down in a wiki-like format for everything to start your own ISP, and what systems/subsystems/technologies you will want to be familiar with, could fill thousands of pages if printed in book form. It's a gargantuan task to make a website that has everything needed to start an ISP from scratch, assuming the reader hasn't been involved in somebody else's ISP operations first.