In IPv6-only networks communication between nodes is just that: IPv6-only. Communication between a node on the IPv6-only network and a remote node reachable only over IPv4 is not possible, because the hosts can only communicate using IPv6 at the network layer. This is where translation techniques come into place, which can operate at many different layers. We categorize translation techniques by the layer they may appear in, that is the network layer, the transport layer or the application layer.

In the network layer the header of a datagram is translated from IPv6 to IPv4 (or vice versa), which happens in the operating system of the originating host. In the transport layer the general mechanism is the use of a relay, which the data has to pass through. This relay is commonly a dual stack device that will translate and pass on datagrams between the different networks. In the application layer an “application layer gateway” (ALG) is used, e.g. a web proxy. While in the network and transport layer translating and relaying IPv4/IPv6 datagrams is mostly application independent, application layer gateways have to be set up for each and every application or service one wants to offer.