IPv4 vs. IPv6
Compare IPv4 and IPv6 on address space, NAT, headers and adoption — and understand why IPv6 proxies behave differently for data collection.
IPv6 was designed to replace IPv4 because the world ran out of IPv4 addresses. Three decades later both run side by side, and the practical differences matter more than the version number suggests.
1. The Address Space
IPv4 uses 32 bits: about 4.3 billion addresses.
192.168.1.1
IPv6 uses 128 bits: about 340 undecillion.
2001:0db8:85a3:0000:0000:8a2e:0370:7334
2001:db8:85a3::8a2e:370:7334 (same, compressed)
The scale difference is hard to convey usefully. The practical consequence: IPv4 addresses are scarce and have market value, IPv6 addresses are effectively free, and a single residential IPv6 allocation typically contains more addresses than the entire IPv4 internet.
2. Side by Side
| IPv4 | IPv6 | |
|---|---|---|
| Address size | 32 bits | 128 bits |
| Notation | Dotted decimal | Colon-separated hex |
| Total addresses | ~4.3 billion | ~3.4 × 10³⁸ |
| NAT | Ubiquitous | Rarely needed |
| Header | Variable, with checksum | Fixed 40 bytes, no checksum |
| Fragmentation | Routers may fragment | Sender only |
| Address config | DHCP or static | SLAAC or DHCPv6 |
| Broadcast | Yes | Replaced by multicast |
| IPsec | Optional | Part of the design |
| Adoption | Universal | Roughly 40–45% of traffic |
3. NAT — The Real Difference in Practice
IPv4 scarcity produced Network Address Translation. Your home network's devices share one public address, with the router translating between them.
NAT has consequences everybody has lived with: inbound connections need port forwarding, peer-to-peer needs hole punching, and — most relevantly here — many users share one public IP.
IPv6 removes the need. Every device can hold a globally routable address.
For anything that identifies by IP, this cuts both ways:
- IPv4: an address may represent one user or ten thousand behind carrier-grade NAT. Blocking one punishes many, which is exactly why sites are cautious about blocking residential IPv4.
- IPv6: an address usually maps to one device, so it identifies far more precisely. But providers hand out enormous blocks, so a single subscriber may hold billions of addresses.
4. Why This Matters for Data Collection
The consequences are specific and often surprising.
IPv6 proxies are cheap and plentiful, because the addresses cost nothing. A provider can offer effectively unlimited unique IPv6 addresses at prices IPv4 cannot approach.
Sites often block IPv6 by prefix, not by address. Since one customer holds a /64 or larger, blocking a single IPv6 address is pointless — rotating within your own allocation is trivial. Detection systems therefore treat the prefix as the unit. Rotating through a million addresses inside one /64 looks, to the target, like one client.
That is the trap: IPv6 rotation feels like enormous diversity and provides almost none, because the entity being tracked is the block you were allocated.
Not everywhere supports it. A meaningful share of sites are IPv4-only, so an IPv6-only client simply cannot reach them. Dual-stack is required for coverage.
Reputation data is thinner. IPv6 abuse databases are less mature than IPv4's, which cuts in both directions — less history to hold against you, and more suspicion of the unfamiliar.
5. Header and Protocol Changes
IPv6's header is fixed at 40 bytes with a small number of fields, where IPv4's is variable with options. Routers do less work per packet.
The checksum is gone entirely — link and transport layers already provide integrity, so recomputing it at every hop was redundant work.
Fragmentation moved to the sender. Routers no longer fragment packets in transit, which removes a class of complexity and a category of attack.
6. Coexistence
Full migration never happened, so both run together:
- Dual stack. Devices hold both addresses and choose per destination. The dominant approach.
- Tunnelling. IPv6 packets carried inside IPv4 across networks that do not support it.
- Translation (NAT64/DNS64). Lets IPv6-only clients reach IPv4-only servers.
Client behaviour matters here: Happy Eyeballs has clients try both and use whichever answers first, which is why an IPv6 problem often shows up as intermittent slowness rather than a clean failure.
7. Real-World Examples
- An IPv6 proxy pool priced at a fraction of IPv4, with far worse effective diversity because it lived in one prefix.
- A block applied to a whole /64, ending a rotation strategy that had cycled through thousands of addresses.
- An IPv6-only client unable to reach a legacy target, resolved by dual stack.
- CGNAT causing collective rate limiting, where thousands of unrelated IPv4 users shared one public address.
- Intermittent latency traced to Happy Eyeballs, retrying a broken IPv6 path on every connection.
8. Summary
IPv4 offers 4.3 billion addresses and relies on NAT to stretch them; IPv6 offers an effectively unlimited space and removes the need for NAT. The header is simpler, the checksum is gone, and fragmentation belongs to the sender.
For data collection the decisive difference is the unit of identity. IPv4 addresses are scarce, often shared, and therefore blocked cautiously. IPv6 addresses are abundant and allocated in huge blocks, so detection tracks the prefix — making large-scale IPv6 rotation look far more diverse than it is.
Both will coexist for a long time. Dual stack is the practical requirement, and IPv6-only reach remains incomplete.