montasaurus

stamina is an opinionated wrapper around tenacity, so either works great

Thank you. How many neighbors are you thinking? It really depends on how many buildings they are in / how far apart they are, and what percentage use you as their provider. At smaller scales, it's much better to have an "in" with the building (HOA, owner, etc.) so you can market more effectively, outweighing the revenue risk of having a small number of customers.

Sonic has a fantastic privacy policy. For comparison, we follow all the same tenets and believe that both contractual and technical safeguards are the best way to ensure our operators do as well. We should definitely be more public about that to non-customers though--both so prospective customers know and to make sure there's public accountability behind it.

I saw a demo of this in Oakland a few months ago. I thought it was super interesting, especially the way they're extending routing concepts. It feels like a very elegant technical solution.

I don't think of it as a question of fortitude as much as opportunity cost. We're looking for operators that would be successful running other types of businesses instead. We want them to choose the ISP business over one that they might be more familiar with. No doubt they could come to understand a transit contract and all the other components, but it's frictions like those that reduce the number of people who decide being an ISP is the right business for them. If you can 1) find customers and 2) provide a great service experience, we want you selling internet! You can spend a lifetime getting good at those two things. We'd rather our operators do that, and we can handle the networking for them.

Getting good terms from an ILEC that you're directly competing with will be difficult. We advocate building your own last mile infrastructure where possible to get around that. Here in SF, we're using fixed wireless to do that.

We use fixed wireless for our last mile (60GHz and 5.8GHz in SF). We haven't been reliant on Local Exchanges. We'd typically be selling to residential end users (single family homes and/or MDUs).

Thanks! Going through Comcast / AT&T isn't the only path to the internet, plenty of other companies sell the backhaul services that we need--Hurricane Electric, Zayo, Cogent, CenturyLink, etc..

Monitoring traffic and throttling it are both network-level interventions that we won't implement or support in a way that can be abused like that. We don't want to expose our operators to the liability of even collecting this type of info.

We monitor your infrastructure (out of band, where possible) and let you know when things go wrong and how to fix it. When the fixing requires physical intervention, we guide you through that. Similarly for installations, where we teach you how to do them and you (or your employees or contractors) perform the physical installation. We'll help you more generally with the business as well, and share advice on how to effectively run an ISP.

Let us know if you speak with them--we're interested in having an existing ISP in the batch.

It makes sense for a wide range of densities. Getting MDUs is really nice, because your marginal cost to serve another customer in the building is really low. Downside is that it may be more competitive. Rural areas present a different business model, where you may have to go long distances and charge enough to support the infrastructure that requires. Single family homes in suburban/urban densities are right in between.

Each of the cases can work, you just need to make sure you're running the right kind of ISP for the market.

It's largely an infrastructure business at the end of the day, and not without its challenges. The real problem is that there are people willing to take on these challenges, but they get stuck as early as the "get an uplink" part of that process.

Just digging into that piece, there are a lot of decisions. Burstable or dedicated? Why are they billing based on 95th percentile? How much bandwidth should I budget per subscriber? The typical uplink quote we've seen has 6 different tables with like 80 different prices on it. If you've bought it before, you know what the tradeoffs are. If you haven't, that's a steep hill to climb just for step 1 of the process.

We deploy fixed wireless here in SF, mainly 60GHz and 5.8GHz. They require line of sight, but something like Baicells could be a good fit for areas where that's more of an issue. It's not just the NIMBYs when you go to hang/bury your own lines, the incumbents will box you out on power poles and generally make things difficult for you.

We've been able to source enough IPv4s for our needs so far, but we don't expect that to last forever. ARIN is cutting off the allocations, but there are still a decent amount available for lease.

When someone resets their router, it clears the auth details and their internet stops working. With DHCP, it'll "just work" when the device comes back up. You just need to be a bit smart about figuring out which device belongs to who. DHCP Option 82 is a tagging mechanism that allows you to do this without creating a hard dependency on the physical address of the customer's router.