Starting an ISP is hard (2016)

I was involved in starting one of America's biggest ISP's back in the early 90's - I won't name which one, but it was #2 for a while - and I can say that things have radically changed from the way they used to be. But, I don't believe that the fact that "its hard" is a reason not to do it - I think we really need to return to the days of Mom&Pop BBS'es and locally-maintained and operated NOC's in neighborhoods that really need it.

Which is why I think its great that we have social movements such as Funk Feuer (https://www.funkfeuer.at) to look at today, point, and say "thank god for the kids!", keeping the original 'net spirit alive.

Starting an ISP doesn't have to be that hard at all. If you want to start one in the country where I live you could do so with very little investment.

All the issues he talks about are due to the fact that there is no real competition in the US and no infrastructure to support competition.

The way it works over here is that there is a (commercially owned) open fiber network which can be used by anyone who wants to, and prices are the same for everyone. This network owns the last mile fiber, a PoP in each neighbourhood and a City-PoP linked to the neighbourhood PoP's.

You want to start a small ISP with very little investment ? Just lease everything from one of the bigger players. Bandwidth, IPTV service, etc. can all just be repackaged and resold under your own brand name and you can work on getting a customer base. You could run this from your basement without buying a single piece of hardware.

Want to spend a little more ? Use your own backbone network and uplink to the internet, but lease all the last-mile stuff from one of the bigger ISP. Go ahead. Want to do everything yourself ? You can rent space in the PoP's and install your own equipment, all you'd need from a 3rd party is to lease the last mile dark fiber.

You can pretty much choose anything from just reselling an existing package to doing everything yourself except owning the physical fiber and PoP your equipment is placed in. IIRC the situation on DSL is similar.

My current ISP started out as one of those that pretty much resold existing packages and currently they are on the other end of the spectrum and run their entire network by themselves including spinning off their own TV provider that now sells IPTV services to other small ISP's.

Net result of all this: at my address I can choose between 13 ISP's on fiber alone.

How do you get to a situation like this ? Very simple. Make sure the government only issues permits for the installation of a broadband network if the network to be built is going to be open to everyone.

This is obviously a response to an earlier post. This seems to happen a lot. Posts inspiring links, inspiring even more branches. It'd be interesting to apply one of those bioinformatics algorithms for computing the phylogenetic tree of DNA to the stream of links featured here on HN.

These guys are really, really redundant, which is excellent, still I am sure my local fibre ISP doesn't have even 1/3 of that. Probably two uplinks, no second power line and one diesel etc. I haven't seen those expensive splicing machines, technicians have used a laptop and some small tool..

Interesting read. It seems to me some of the long tail risks mitigated by expensive redundancies might be an acceptable risk for the customers. For example, the article indicates that they have redundant power suppliers, along with 2 days worth of diesel for on-premise generation. Apparently keeping the second power company available is a recurring cost. What are the chances power is cut to the office for more than 2 days that (1) doesn't also affect the other power company (e.g., earthquake, hurricane, etc.) and (2) couldn't be mitigated by purchasing more diesel? Even if (1) and (2) are true, couldn't this just be a reasonable outage?

This is a unique perspective on the control Google has on the web. You basically can't start an ISP without access to Google, and access to Google is expensive. And you can't cache/proxy it yourself, because https is, like, really useful, man.

I don't know if I agree with this. I cofounded an ISP in the mid-90's. It was in a relatively small town in a rural state in the midwestern US.

I have no idea what the residents options for internet access are now, but at the time the only really choice (unless you were college faculty or a student) was AOL. But there wasn't a local line - you had to call a city 30 miles away that had a long distance toll charge.

That meant we had a relatively captive audience. But, not a lot of people knew what the "internet" was then.

At the time, our largest cost was the 32 phone lines coming in (more expensive than a residential phone line). Initial equipment for the T1 and the modems (sixteen 28.8 ciscos) and setup was only something like $30k. The second largest cost was the T1 (none of us took a salary).

The issue was capitalization. It got to a point where additional significant personal guarantees were going to be required from people who hadn't had a salary for almost 2 years (I was making a little money making static websites). At that time the state telecom called and offered a little money for the customers and to assist in migrating them (I guess they were finally ready to offer rural internet access).

At the end of the day, we might have been a little early to the market, we were under-capitalized, and we didn't realize at the time that if we had offered complementary services around residential internet access (we only had a couple commercial customers, more aggressive web development and hosting services, and a better idea in how to raise capital without more personal guarantees), who knows.

I should add that I never wanted to be in the ISP business, I had an idea about a web based business (that I still think was a really good idea and I'm sorry that running an ISP and burning through all of our collective savings prevented me from doing) and was talked in to the ISP model by the people I went to looking for investment.

TLDR: Technically, starting and running an ISP isn't all that hard. Probably easier today than it was 22 years ago. The business/economic side is the challenge.

Back when I was 16 and the Dial-up was slowly dying, I took a cable connection and start selling internet to my neighbours.

By the 3rd year the network reached 3 blocks around the neighbourhood and I had ~50 paying clients. I still remember I did that with an old Intel 486 running Slackware Linux and a bunch of ipchains MASQ scripts. I had even a small local website where you were able to see your traffic ( I remember I was paid 10EUR/500MB or something. )

Back then I was not alone. Almost every neighbourhood had those networks, but at one point they started to unite together and the bigger ones were buying the small ones.

Now the biggest problem would be regulation, rather than the hardware requirements.

A nationwide FTTH ISP is a completely different affair to a local WISP...

In the 90s there were ton of mom and pop ISPs. But in the 90s they piggy backed off of telephone lines (dial up) so they didn't have to have an infrastructure aside from running a high speed connection to their office and having a bank of modems connected to phone lines.

I remember there was even a company that let you "start your own ISP" with just a 10 minute sign up form. You used their high speed connection and their modem bank and they took 3/4 the profit.

the original impetus for 'build your own ISP' started from a HN discussion, and frankly i think its where it should end.

Detroit is a sterling reminder of the hubris of a generation in the wake of late stage capitalism. Naturally, creating your own internet in the wasteland that was once americas crown jewel of middle class prosperity seems not only natural, but necessary as the economic incentive from private industry is nonexistent.

For the rest of us, effort is best spent in cultivating a healthy local network. build your own open source router, control traffic according to your own best practices. For example, limit advertising servers and telimetry servers so as to diminish the unspoken incentive the repeal of net neutrality and operating systems that do not respect privacy. Host your own VPN for security and DNS server to avoid SRVFAIL hijacking.

https://prism-break.org http://www.pcengines.ch/alix.htm

Indeed, it sounds easier to start a bank. Probably more profitable also.