IPv4 vs IPv6 is a comparison between two versions of the Internet Protocol, the system that gives every online device an address so packets know where to go. IPv4 is the original design. It can hand out about 4.3 billion addresses, and that supply ran out years ago. IPv6 is its successor, built with a supply so large that running out stops being a concern.
This guide explains how each one works, including NAT, the workaround that kept IPv4 alive, and SLAAC, the way IPv6 devices set up their own addresses. It then compares the two on speed, cost, ease and scale, and ends with a simple rule for which one to run.
- IPv4 uses 32-bit addresses, about 4.3 billion in total, and relies on NAT to share them.
- IPv6 uses 128-bit addresses, a practically unlimited supply, so every device can have its own address.
- SLAAC lets IPv6 devices configure their own addresses without a DHCP server.
- Speed is a draw; cost and scale favour IPv6, while familiarity and legacy support favour IPv4.
- Rule of thumb: run dual-stack today, grow on IPv6, and keep IPv4 for compatibility.
- 0:00Intro
- 0:26Two address systems, one internet
- 1:23How an IPv4 address works
- 2:19NAT: many devices, one address
- 3:19IPv6 gives every device an address
- 4:19SLAAC: addresses that set themselves up
- 5:09Is IPv6 faster?
- 6:00Cost versus familiarity
- 6:49Scale versus legacy support
- 7:45You're building a new office network
- 8:37The rule of thumb
- 9:33What to remember
- 10:09IPv6 first, IPv4 for compatibility
IPv4 vs IPv6: what is the difference?
Both IPv4 and IPv6 do the same job. They are versions of the Internet Protocol, the layer that addresses and routes every packet you send. The difference is in the design. IPv4 is the 1981 standard, with 32-bit addresses and about 4.3 billion in total. IPv6 is the successor, with 128-bit addresses, a practically unlimited supply, and automatic setup through SLAAC.
The comparison is trickier than it first looks. IPv6 fixes the address shortage, but decades of software, devices and habits were built around IPv4. That means the newer option is not automatically the easier choice.
Both run today, and many networks carry them side by side. Your phone may already use IPv6 for some sites and IPv4 for others without you noticing. So the useful question is not which protocol wins outright, but how to mix them well.
- IPv4: 1981 standard, 32-bit, about 4.3 billion addresses, stretched with NAT
- IPv6: successor, 128-bit, practically unlimited, end-to-end, plus SLAAC
How does an IPv4 address work?
An IPv4 address is four numbers, each stored in eight bits. Take a typical home address: 192.168.1.10. That dotted decimal form is what people write. The computer sees it in binary, with each number turned into eight bits: 11000000 10101000 00000001 00001010. Four groups of eight bits make 32 bits in total.
Thirty-two bits can only count so high. All possible addresses together come to two to the power of 32, roughly 4.3 billion. When IPv4 was written, the internet was a research network, and a few billion addresses looked like far more than anyone would ever need.
That assumption did not last. Phones, laptops and TVs each want an address, a single household can have a dozen connected devices, and there are billions of households. Plain IPv4 simply cannot give everyone their own address.
What is NAT, and why does IPv4 need it?
NAT stands for Network Address Translation. It is the workaround that kept IPv4 working long after the addresses ran out, and you are almost certainly behind one right now. Your router hides a whole home behind a single public address.
Here is how it plays out. Your laptop at 192.168.1.10 and your phone at 192.168.1.11 both send traffic through the NAT router. The router swaps each private address for its own public one, say 203.0.113.5, and sends the packet across the internet. When the reply comes back, it swaps the address back again. The web server only ever sees that one public address.
The cost is complexity. The router has to track every connection in a translation table and juggle ports, and outside devices cannot easily reach yours directly. In short, NAT saves addresses but costs simplicity.
How IPv6 gives every device its own address
IPv6 does not stretch the old supply. It replaces it with a far bigger one. Its addresses are 128 bits, four times the bits of IPv4, which multiplies the supply so much that running out stops being a concern.
Picture the same home with the same devices. Now the laptop has its own global address, and so does the phone. The home router still routes and filters traffic, but it does not rewrite anything. Packets cross the internet unchanged, so the web server sees exactly which device is talking.
This end-to-end design means apps such as calls and games can connect directly, without translation in the way. You still use a firewall for safety, but each address stays true to its device. IPv6 also uses a fixed-size base header, which is easier for routers to process.
- 128-bit addresses: supply practically unlimited
- End-to-end: devices reach each other directly
- Simpler headers: a fixed base header routers handle cleanly
What is SLAAC in IPv6?
With addresses this long, nobody wants to type them in, and nobody has to. SLAAC, or Stateless Address Autoconfiguration, lets devices set up their own address automatically. The router shares a network prefix, and the device does the rest. "Stateless" means the router does not keep a list of which device got which address.
Watch a laptop join the network. First it sends a Router Solicitation, asking whether there is a router present. The router replies with a Router Advertisement (RA) carrying a prefix such as 2001:db8:1::/64. The laptop adds its own ending, then runs duplicate address detection (DAD), asking whether ::a1b2 is already taken.
If nobody answers that check, the address is unique and the laptop starts using it. No DHCP server is needed to hand out addresses. You plug in and IPv6 just works.
- Laptop sends a Router Solicitation
- Router answers with a Router Advertisement and prefix
- Laptop builds an address and checks it with DAD
- Silence means the address is unique
Is IPv6 faster than IPv4?
Speed is usually the first thing people ask about, and the honest answer is that it is a draw. In practice, throughput is similar on both. IPv4 headers vary in size and can carry options, while IPv6 uses a fixed base header. IPv4 traffic often passes through a NAT translation, and IPv6 does not need one. But for both, what really decides speed is the network path.
If your provider sends IPv6 traffic a longer way round, it can feel slower, and sometimes the reverse is true. A shorter route beats a newer protocol, and the protocol itself is rarely the bottleneck. So do not switch to IPv6 expecting more speed, and do not avoid it fearing it will be slow. Base the decision on cost and scale instead.
Cost vs ease: scarce addresses or familiar tools
Here the two options pull in opposite directions. IPv4 addresses are scarce, so public ones cost money. Most admins know IPv4 well, though NAT adds hidden work. IPv6 addresses are plentiful, so nothing needs rationing, but many teams find IPv6 less familiar, with long addresses and new habits to learn.
On cost, IPv6 wins. With IPv4, every new server or customer means finding another scarce public address. With IPv6, that supply problem simply goes away.
On ease, IPv4 wins, at least for now, because people already know how to read and debug it. IPv6's main cost is not money but learning, and that cost shrinks over time as teams get used to it.
- Cost round: IPv6, no paying for scarce public addresses
- Ease round: IPv4, familiar tools, habits and troubleshooting
Scale, legacy support and a real office network
The raw numbers explain the scale round. About 4.3 billion is the entire IPv4 space, fewer than one address per person on Earth before you count servers and devices. IPv6 offers two to the 128, enough that every device you will ever own can have its own address. As you add devices, sites or customers, you simply add addresses, with no translation layers stacked on top. Scale goes to IPv6.
Legacy support goes to IPv4. Some older devices, apps and services only speak IPv4, and it still offers the broadest compatibility. IPv6 is where things are heading, but IPv4 remains the bridge to everything older.
Consider a growing office. Staff laptops get both kinds of address and send traffic to a dual-stack router. Anything that can go native takes the IPv6 path, set up by SLAAC, out to modern sites. Anything older goes through IPv4 NAT, sharing one public address, to reach something like an older supplier portal. Going IPv6-only would break that portal; staying IPv4-only would add to the NAT burden with every new device.
- Scale round: IPv6
- Support round: IPv4
- Office pick: dual-stack, IPv6 for growth, IPv4 for reach
IPv4 or IPv6: which should you pick?
You do not have to pick one protocol forever. You have to give each one the right role. The rule of thumb: run dual-stack today, prefer IPv6 for growth, and keep IPv4 for compatibility. IPv4 covers the older devices and services that still need it. IPv6 covers every network, device and customer you add from here on. Where they overlap is dual-stack, and that is where nearly everyone should be today.
Pick IPv6 as the default if you are building something new, growing fast, or tired of buying scarce IPv4 addresses. Keep IPv4 if any device, app or partner you rely on only speaks IPv4, but treat it as a compatibility layer, not the place you grow. Dual-stack is not sitting on the fence. It means IPv6 first, with IPv4 kept around until nothing needs it.
- Pick IPv6 if you're building new, scaling, or paying for addresses
- Keep IPv4 if something you depend on only speaks IPv4
- Run both until IPv4 is truly unneeded
Key takeaways
- IPv4 uses 32 bits, about 4.3 billion addresses, and the supply has run out.
- NAT shares one public address across many devices, which saves addresses but adds complexity.
- IPv6 uses 128 bits, a practically unlimited supply, so devices connect end to end.
- SLAAC lets IPv6 devices configure their own addresses without DHCP.
- Speed is a draw; cost and scale favour IPv6, familiarity and legacy support favour IPv4.
- Run dual-stack and grow on IPv6.
Frequently asked questions
Why did IPv4 run out of addresses?
IPv4 addresses are 32 bits long, which allows only about 4.3 billion of them. It was designed in 1981, when the internet was a research network, long before billions of households each had a dozen connected devices.
Does IPv6 still use NAT?
IPv6 does not need NAT, because there are enough addresses for every device to have its own. Your router still routes and filters traffic, and you still use a firewall, but packets cross the internet without being rewritten.
Is IPv6 faster than IPv4?
Not in any reliable way. Throughput is similar in practice, and the network path your provider uses matters far more than the protocol. Choose based on cost and scale rather than speed.
Do I need DHCP with IPv6?
Not necessarily. With SLAAC, the router advertises a network prefix, the device builds its own address and checks that no one else is using it. No server needs to hand out addresses.
Should I switch my network to IPv6-only?
For most networks, not yet. Some older devices, apps and services only speak IPv4 and would break. Running dual-stack lets you grow on IPv6 while keeping IPv4 for compatibility.
What does dual-stack mean?
Dual-stack means devices and routers run IPv4 and IPv6 at the same time. Traffic that can use IPv6 does, and anything older falls back to IPv4, usually through NAT.
