Start Your Own ISP

Original Article ↗ Hacker News Discussion ↗

Regulation and feasibility by region

  • Mixed views on Canada: some describe heavy regulation, regulatory capture, and hurdles around wholesale access (especially fiber), citing small ISPs being bought by incumbents; others say barriers are low, especially for rural WISPs.
  • Mention of a Canadian government move to require telecoms to provide backup for each other after a major outage, but details and impact are debated and somewhat unclear.
  • Similar complaints from Spain and San Francisco about municipal barriers and entrenched incumbents.
  • In the US, some counties have municipal fiber constraints and cost-sharing models that make individual upgrades expensive.

Economics and business risk

  • Classic joke: “how to make a small fortune with an ISP: start with a large one.”
  • Some report starting small WISPs with modest funds, but note that profitability is hard and regulatory/wholesale dynamics can force small ISPs to sell out.
  • Fiber builds face high upfront capex and permitting delays; some operators focus on MDUs and dense areas to make the economics work.

Starlink, 4G/5G, and competitive pressure

  • Many see Starlink as a major threat for rural WISPs, noting customers defecting from WISPs and DSL.
  • Counterpoints:
    • Starlink is more expensive than cable/fiber, often slower, and capacity-limited per area.
    • It’s seen as a replacement for dial-up, legacy satellite, and weak DSL, not for good cable/fiber.
    • Satellites have limited lifetimes but the constellation is continuously refreshed.
  • 4G/5G fixed wireless: experiences range from “budget, high-latency, CGNAT” to “surprisingly good if capacity-managed,” especially for low-end users.

Technical practices and equipment

  • Good WISPs can outperform Starlink, especially if they:
    • Pull fiber near communities and distribute via 60 GHz or 5 GHz.
    • Gradually reinvest profits to extend fiber deeper.
    • Serve MDUs efficiently.
  • Reality described as “cowboy” operations: poor design, technical debt, overuse of certain low-cost gear, and reluctance to pay real network engineers.
  • Debate over gear:
    • Low-cost platforms (e.g., Mikrotik, Ubiquiti) are great for homelabs and small setups but can struggle at ISP scale, especially with advanced routing/MPLS.
    • Higher-end vendors (Juniper, Nokia, etc.) are recommended for core/edge in serious deployments.

Addressing and protocol issues

  • IPv4 address costs are a significant barrier; ARIN waitlists and auctions are mentioned.
  • Some ISPs run IPv6-first with NAT64/DNS64 or similar to minimize IPv4 needs.
  • CGNAT on cellular/FWA causes connection instability and is a pain for advanced users.

Quality of advice and resources

  • The “start your own ISP” site is seen as a useful high-level starting point but not authoritative.
  • Concerns raised about relying on random online guides and Facebook groups; good network design advice is hard to identify and usually costs money.
  • Several talks and case studies of individuals building fiber ISPs are recommended as more in-depth resources.