Static, Dynamic, Sticky and CGNAT: How Addressing Actually Works

Why dynamic is the default: the DHCP lease

Addresses on a consumer line are leased, not given. RFC 2131 defines the period an address is allocated to a client as a lease, and states that the mechanism reuses addresses whose lease has expired. A client renews at 50% of the lease (T1) and rebinds at 87.5% (T2). Addresses are occupied only while a line is active, then recycled.

This is pure economics for the ISP. IPv4 exhausted at the IANA central pool on 3 February 2011 and at the regional registries over 2011–2015, and addresses now trade as a commodity — roughly $20–30 each on the mid-2026 market. A permanent address per household wastes a resource the ISP can otherwise reuse or sell, which is why static is a paid, business-tier feature.

CGNAT: you may have no public address at all

Carrier-grade NAT, standardised in RFC 6598, assigns subscribers addresses from the shared 100.64.0.0/10 block and shares a single public address among many of them at the carrier level. The /10 was defined precisely so ISPs could deploy CGN regionally without the address conflicts that reusing RFC 1918 private space would cause.

Behind CGNAT, no unique public address is assigned to your line, so inbound connections and port forwarding fail at the carrier gateway — you cannot host a reachable service. You can check: if your router's WAN address falls in 100.64.0.0–100.127.255.255, or differs from the address the internet sees, you are behind it. Settle this before paying for any static plan: a static address only helps if the carrier takes your line off CGNAT along with it.

Static vs sticky vs dynamic vs DHCP reservation

Sticky is not a standardised term — there is no RFC for it. It describes a dynamic address that persists only because the lease keeps renewing while the line stays up, exactly the RFC 2131 mechanism above. It has no guarantee behind it, which is why a reinstall or a carrier-side change can return a different address, and why sticky sold as static is a real problem.

Why home upload is slower than download

On cable (DOCSIS) the cause is physical spectrum. The traditional split gives upstream only 5–42 MHz against downstream from 54 MHz up to around 1 GHz — the upstream band is a fraction of the downstream, because the HFC network was originally built to broadcast television one way. Newer splits widen upstream to 85 or 204 MHz but the asymmetry remains.

Fibre (FTTH) is physically capable of symmetric speeds; there the asymmetry is a commercial choice, not a physical limit. Either way, on a residential line the upload figure is the one that caps what a hosted service can deliver, and it is shared across everyone on the line.

Who controls the reverse record (PTR)

The reverse DNS record is controlled by whoever owns or allocates the IP block — the ISP — not by the holder of any domain name. Consumer dynamic ranges get bulk auto-generated PTRs the user cannot change; business and static ranges can usually request a custom PTR through the ISP.

This has a concrete consequence: Spamhaus's Policy Block List explicitly lists IP space that should not send email directly — typically ISP broadband ranges assigned to consumer customers. A residential address is low-trust to mail systems by policy, not by accident, and that is a property of where the address sits, not of your server.