Kelly Mears

IPv6

IPv6 exists mainly to end address scarcity, with a large enough space that per-device NAT becomes optional.

Networks & Distribution2 min read290 words12 out · 12 in
also calledInternet Protocol version 6

IPv6 is the successor to IPv4, and its headline change is address space: 128 bits instead of 32, which is the difference between roughly 4 billion addresses and roughly 340 undecillion. IPv4 exhaustion was predictable decades in advance and IPv6 was designed to solve it outright rather than patch around it, but adoption has been slow precisely because NAT Traversal and address-sharing techniques kept IPv4 usable well past its intended lifespan.

The space is large enough to change the deployment model, not just extend it. Under IPv4, home networks share one public address across many devices via NAT, and that translation layer is why NAT Traversal is a whole discipline in the first place. Under IPv6, every device can plausibly get its own globally routable address, which removes the NAT layer entirely — a return to the original end-to-end internet model where any two hosts can address each other directly, no relay or hole-punching required. This is also why some engineers view widespread IPv6 as slowly obsoleting parts of the Peer-to-Peer and WebRTC connectivity toolchain that exists specifically to work around NAT.

IPv6 addresses look different enough to trip people up: eight groups of four hex digits separated by colons, with runs of zeros collapsible to :: once per address. Autoconfiguration (SLAAC) lets a device generate its own address from the network prefix and its interface identifier without a DHCP server, which is a meaningful operational simplification at scale.

Dual-stack — running IPv4 and IPv6 side by side — has been the transition strategy for over a decade, and it's the reason both protocols still need to be understood: a client and server can each prefer a different one, and "happy eyeballs" algorithms race both to pick whichever connects first.

See also9

Anycast

Anycast announces the same IP address from many locations and lets normal routing send each client to the nearest one.

Networks & Distribution8 connections

TCP and UDP

TCP guarantees ordered, reliable delivery at the cost of latency; UDP delivers packets fast with no such promise.

Networks & Distribution17 connections

Domain Name System

The distributed directory translating names to addresses, and the layer where propagation delays live.

Systems & Tooling27 connections

Zero Trust Network

Zero trust authenticates and authorizes every request individually, treating network location as no proof of anything.

Networks & Distribution12 connections

Ethernet

The wired local networking standard that carries most of the world's traffic over its first hop, from the frame format and MAC addresses to switches, ARP, and power over the same cable.

Networks & Distribution9 connections

Port Scanning

Probing a machine's network ports to find which services are listening, the first step of both attacks and audits, now done to the entire internet in minutes, so anything exposed gets found.

Networks & Distribution9 connections

Loopback

The network address a computer uses to talk to itself, where packets go out and come straight back without touching any hardware, the home of development servers and a recurring blind spot in security.

Networks & Distribution12 connections

Network Address Translation

Rewriting the addresses in packets as they cross a router so that many devices can share one public IPv4 address, a stopgap that became permanent and broke the internet's end-to-end design.

Networks & Distribution17 connections

Subnet

A block of IP addresses that share a common prefix, written in CIDR notation such as 192.168.1.0/24, which decides what a device can reach directly and what it must send through a router.

Networks & Distribution13 connections

Linked from12