IPV6 FOR THE CCNA:
COMPRESSION, ADDRESS TYPES & EUI-64

IPv6 is not IPv4 with more bits. It's a different addressing model, and most of the friction CCNA students hit comes from trying to drag IPv4 habits into IPv6 — looking for a "broadcast", asking for the "first usable host", wondering where the subnet mask went. None of those translate cleanly.

The good news: there's less to memorize than you'd think. Three concepts — compression, address types, and EUI-64 — cover most of what the exam tests. This article walks each one end-to-end, then points at a free browser tool so you can drill until it sticks.

If you've read the subnetting article and worked through the IPv4 practice tool, you're already past the hardest part. IPv6's math is actually simpler — there are no host counts to subtract two from, no broadcast addresses to compute, and the prefix is always written in slash notation.

★ 128 BITS, 8 HEXTETS

An IPv6 address is 128 bits, written as 8 groups of 4 hexadecimal characters separated by colons:

Each group is called a hextet (some textbooks say "quartet" or "group"; "hextet" is the modern term). Each hex character is 4 bits, so each hextet is 16 bits, and 8 × 16 = 128. The address space is 2¹²⁸ — enough that the math gets silly fast, which is the whole point.

That fully-expanded form is correct, but no one writes addresses like that in production. There are two compression rules that collapse it down, and the exam will ask you to apply both.

★ COMPRESSION — TWO RULES, ONE GOTCHA

Applied to our example:

The gotcha: when there are two equal-length runs of zeros, you compress the first one by convention (RFC 5952). Most compressors get this right; exam questions sometimes pick addresses where the rule actually matters.

★ THE 6 ADDRESS TYPES YOU MUST RECOGNIZE

IPv6 has more address types than IPv4, but only six matter for the CCNA. Memorize the leading bits of each and you can identify any address at a glance.

No broadcast. IPv6 doesn't have a broadcast address. Anything that would have been broadcast in IPv4 — ARP, "all hosts on the subnet" — is now a multicast group. ff02::1 reaches every node on the link; that replaces broadcast for most practical purposes.

The Link-Local one is the most under-appreciated. Every IPv6 interface, the moment it comes up, generates an fe80:: address using either EUI-64 or a random interface ID. OSPFv3 and EIGRP for IPv6 form their neighbor adjacencies over link-local. So even if you never configure a global address, IPv6 is doing things on that interface.

★ EUI-64 — WHEN A MAC BECOMES AN INTERFACE ID

Modified EUI-64 is how a router can generate the lower 64 bits of an IPv6 address from a MAC address without manual configuration. The exam tests the transform in detail. There are three steps and one bit-flip that trips everyone up the first time.

The U/L bit (Universal/Local) is bit 1 of the first byte. If it's 0, the MAC is universally administered (burnt-in). If it's 1, it's locally administered. IPv6 inverts it for EUI-64 — the spec wants 1 to mean "universally unique" in the interface ID, so universally administered MACs (which have U/L = 0) get flipped to 1.

Why memorize this? Because exams test it, and because when you're troubleshooting and you see an fe80:: address with an FF:FE in the middle of it, you can read the original MAC right off — flip the U/L bit back, drop the FF:FE, and you have the burnt-in MAC. Useful for matching a Layer 3 trace to a Layer 2 port.

Reformatted into IPv6 hextets, the interface ID from our example becomes 020c:29ff:feab:cdef. Combined with an fe80::/64 prefix, the link-local address is fe80::20c:29ff:feab:cdef.

★ /48, /56, /64 — THE SUBNET HIERARCHY

IPv6 subnetting is much simpler than IPv4 because you almost always work at three prefix lengths:

A /48 gives you 2¹⁶ = 65,536 /64 subnets. That sounds excessive — and it is, by design. IPv6 trades address efficiency for routing simplicity. If you have a /48 and you need to label subnets by floor, by VLAN, by department, you have all the bits you need.

There is no VLSM in IPv6 the way there is in IPv4. Yes, you can technically use other prefix lengths — and routers will route them — but every protocol that depends on a 64-bit interface ID (SLAAC, EUI-64, neighbor discovery, OSPFv3's link-local adjacency) assumes /64. Going smaller breaks them. Go bigger only when you're carving up a /48 into /56s or /60s for sub-allocations, never on the host segment.

For the exam, if a question says "subnet this /48 into /64s, give me the 7th subnet," you increment the 4th hextet by 6 (since the first is the original, the second is +1, etc.). That's it. No host arithmetic, no broadcast to skip.

★ SUGGESTED STUDY ORDER

The whole loop is maybe 90 minutes of focused work. The reason it's fast is that IPv6 has fewer arbitrary rules than IPv4 — once you know the leading-bit table and the EUI-64 transform, there isn't much else to forget under exam pressure.

★ WHAT THIS ARTICLE INTENTIONALLY SKIPS

For the CCNA, the above is enough. There are several IPv6 topics this article skips on purpose, because they don't appear on the 200-301 blueprint or only show up in deeper sub-objectives:

The fastest way to get to "ready for IPv6 questions on the exam" is to nail compression, types, and EUI-64 cold — then sweep the other topics for 20 minutes. They build on the foundation, not around it.