| IPv6-to-IPv6 Network Prefix Translation | |
| Purpose: | Network address translation |
| Rfcs: | ... |
IPv6-to-IPv6 Network Prefix Translation (NPTv6) is a specification for IPv6 to achieve address-independence at the network edge, similar to network address translation (NAT) in Internet Protocol version 4 (IPv4). It has fewer architectural problems than IPv4 NAT; for example, it is stateless, uses a 1:1 address mapping and preserves the reachability attributed to the end-to-end principle. However, because the new address is chosen in a way that leaves the checksum unchanged (checksum-neutral mapping) the interface identifier bits change and this may break applications that embed data in them (such as IPsec). Additionally, split-horizon DNS may be required for use in a business environment.
NAT66 was the name used in earlier drafts of the standard.[1] There were some initial proposals to rename it,[2] and a few years later the name NPTv6 was chosen.[3] One of the early versions defined two modes of operation within NAT66: a Two-Way Algorithmic mapping and a Topology Hiding Option, the latter of which used a non-reversible address mapping that would've required additional state in the translator, either in the form of a dynamic table or a statically defined set of address mappings.[4] It was soon removed, leaving the two-way mapping as the only mode of operation and making the specification fully stateless.
Current usage of the term (by vendors, in informal contexts, etc.) is unclear: sometimes it's still employed as a synonym for NPTv6[5] but often it refers to a generic implementation of stateful NAT[6] [7] [8] (or even full NAPT[9] [10]) on IPv6.
