Field Guide — Networking Fundamentals
IPv6 Address Analyzer & Scope Tool
Paste any IPv6 address. Get its canonical form, scope, RFC reference, and the operational story behind why that scope exists — for engineers who care, and stakeholders who just need the answer.
Analyze an IPv6 address
Accepts shorthand (::), zone identifiers (fe80::1%eth0),
bracketed URL form ([::1]), and IPv4-mapped (::ffff:192.0.2.1).
Press Enter to analyze.
The IPv6 address landscape
IPv6 abandoned the classful A/B/C/D model entirely. Instead, every address belongs to one of a small set of scopes — loopback, link-local, unique local, multicast, or global unicast — and the scope is encoded directly in the leading bits. Click any card to load it into the analyzer.
::1/128
Loopback
The IPv6 localhost. Equivalent to 127.0.0.1.
::1::/128
Unspecified
No address — used as source during SLAAC and DHCPv6 solicit.
::fe80::/10
Link-local
Auto-configured on every interface. Used by Neighbor Discovery. Never routed.
fe80::1fc00::/7
Unique Local (ULA)
Private, internal. The IPv6 analog of RFC1918. In practice fd00::/8.
fd12:3456:789a::1ff00::/8
Multicast
Replaces IPv4 broadcast. Scope encoded in the fourth nibble.
ff02::12000::/3
Global Unicast
Publicly routable space allocated by RIRs. Most production traffic.
2607:f8b0:4005:80a::200e2001:db8::/32
Documentation
Reserved for examples. Must never be routed publicly.
2001:db8::12002::/16
6to4 (deprecated)
Legacy transition tech. RFC 7526 deprecated 6to4 anycast in 2015.
2002:c000:0204::1::ffff:0:0/96
IPv4-mapped
Embeds an IPv4 address. Socket-API convention only — never on the wire.
::ffff:192.0.2.1Anatomy of an IPv6 address
An IPv6 address is 128 bits, written as eight groups of four hexadecimal digits separated by colons. The leading bits identify the global routing prefix (assigned by your RIR or ISP); the middle bits are your subnet ID; the trailing 64 bits identify the interface itself. The split — 64 bits of network, 64 bits of host — is fixed for nearly every operational LAN.
In plain English
IPv6 gives every enterprise enough address space to never run out. A single /48 allocation — the standard ISP handout — contains more addresses than the entire IPv4 internet, multiplied by itself seventy-nine quintillion times. The scarcity that defined IPv4 operations for thirty years is gone.
For the engineer
The 64/64 prefix-host split is not arbitrary — it is required by
SLAAC, which derives the interface ID from the MAC address via EUI-64
or generates a random 64-bit value (RFC 7217). Subnetting a /64 into
smaller pieces breaks NDP and Duplicate Address Detection. If you ever
see /127 in production, it is on a point-to-point link
— the one sanctioned exception (RFC 6164).
How scope is encoded in the address itself
Where IPv4 made you remember that 10.0.0.0/8 is private and
224.0.0.0/4 is multicast, IPv6 puts the scope directly into
the leading bits. You can identify the scope of any IPv6 address by reading
its first nibble. This is the single biggest operational improvement over
IPv4 — there is no ambiguity about whether a packet should leave a
link.
fe80::1, then
fd12:3456:789a::1, then 2607:f8b0:4005:80a::200e.
Same first nibble → same scope → same operational meaning. The
analyzer will surface the matching RFC and the engineering callout for
each.
Multicast scope reference
Inside the multicast prefix ff00::/8, the fourth nibble
encodes scope. This is structural, not by convention — firewalls
and routers honor the scope flag without additional configuration.
| Scope nibble | Prefix | Scope name | Reaches |
|---|---|---|---|
1 | ff01::/16 | Interface-local | Same interface (loopback only) |
2 | ff02::/16 | Link-local | Same Layer 2 segment |
4 | ff04::/16 | Admin-local | Manually scoped boundary |
5 | ff05::/16 | Site-local | Within a site (e.g. campus) |
8 | ff08::/16 | Organization-local | Within an org (multi-site) |
e | ff0e::/16 | Global | Internet-wide multicast |
This will bite you
IPv6 has no broadcast. Firewalls and ACLs copied from IPv4 will
silently drop Neighbor Discovery (which uses ff02::1:ff00:0/104
solicited-node multicast) and break dual-stack connectivity in ways
that look like physical-layer problems. Always permit ff02::/16
traffic on every interior interface.
How an interface gets an address (SLAAC)
IPv6 hosts auto-configure addresses without DHCP via Stateless Address Autoconfiguration (SLAAC). The host generates a link-local address from its MAC, sends a Router Solicitation, and uses the prefix advertised by the router to build a global address.
For the engineer
Modern stacks have largely abandoned EUI-64 in favor of RFC 7217 stable-privacy addresses — same /64 prefix, but the host portion is a hash of the interface and a secret. This prevents the MAC-tracking privacy leak that defined early IPv6 deployments. Windows and recent Linux ship RFC 7217 by default; macOS uses RFC 8981 temporary addresses on top.
Generating Unique Local Addresses
Unlike IPv4 RFC1918 — where everyone uses the same handful of
private ranges — ULA prefixes are designed to be statistically
unique. The 40 bits of Global ID after the fd00::/8 prefix
should be randomly generated, so that any two organizations have an
effectively zero collision probability during an M&A merger.
Generate a sanctioned ULA prefix
On Linux: printf 'fd%02x:%04x:%04x::/48\n' $((RANDOM & 255)) $RANDOM $RANDOM
— or use any RFC 4193 generator. Record the prefix in IPAM the
same way you would track an RFC1918 supernet. Never sequence ULAs
(fd00::, fd01::, fd02::) — that defeats the entire
point of the design.
This will bite you
A surprising number of vendor "default" ULA configurations ship with
fd00::/8 hard-coded as the prefix. When two such networks
merge, every host conflicts. Before approving any vendor default,
verify the ULA Global ID is actually random — or override it
with one you generated yourself.
What IPv6 actually changes for operators
IPv6 is not "IPv4 with more addresses." The address space is the headline, but the operational shifts run deeper. Five changes matter most in day-to-day infrastructure work.
1. No broadcast, no ARP
Neighbor Discovery Protocol replaces ARP using link-local multicast.
Broadcast storms become impossible; ARP-cache pollution disappears.
The trade-off: every interior firewall must permit ff02::/16.
2. SLAAC instead of (or alongside) DHCP
Hosts can configure themselves without any server. DHCPv6 still exists for stateful assignment and option distribution, but the network can function without it. This eliminates the single-point-of-failure that DHCP servers represent in IPv4.
3. No NAT in the default path
Every host gets a globally routable address. NAT44/PAT goes away, along with its symmetry breaking, port-exhaustion failures, and ALG kludges. End-to-end addressing returns. Some operators still deploy NPTv6 (RFC 6296) for renumbering convenience — that is their choice, not an architectural requirement.
4. Mandatory link-local on every interface
Every interface auto-configures fe80::/10 at boot. This
is the foundation for NDP, RA, and DHCPv6 messaging. It is also the
reason interface zones (%eth0) exist — the same
fe80::1 may be present on every interface of a host.
5. Fragmentation is host-only
Routers no longer fragment IPv6 packets. Path MTU Discovery becomes mandatory, and PMTUD-blocking firewalls (a common IPv4 misconfiguration) become an immediate dual-stack failure mode. Audit ICMPv6 Type 2 ("Packet Too Big") permits before any rollout.
Working with IPv4?
The IPv4 sibling tool covers the classful A/B/C/D/E model, CIDR notation, RFC1918 private space, APIPA, CGNAT, and reserved ranges — with the same analyzer pattern.