In this section we describe methods for carrying IPv6 over existing IPv4 networks, which invariably means using some kind of tunnelling mechanism, either manually or automatically configured.

Configured tunnelling is defined in the update to RFC 2893 [NG05] as IPv6-over-IPv4 tunnelling where the IPv4 tunnel’s endpoint address is determined by configuration information on the encapsulating node. Therefore the encapsulating node must keep information about all the tunnel endpoint addresses. These kinds of tunnels are point-to-point and need to be configured manually.For control of the tunnel paths, and to reduce the potential for tunnel relay denial-of-service attacks, manually configured tunnels can be advantageous over automatically configured tunnels.

Configured tunnels are best employed when providing external IPv6 connectivity to a whole network. There are not yet many providers who offer IPv6 in any way but if one has the possibility to get initial IPv6 connectivity from an already IPv6-connected site, one of the easiest, most stable and secure ways to get the IPv6 traffic routed to the (as yet) unconnected site is via an IPv6-in-IPv4 tunnel. One can even set up a BGP peering on that link, although if the tunnel is the only off-site link, BGP is not required unless the connecting site is interested in seeing the BGP routing information of its upstream provider. Usually static routes meet the requirements.

Within a site, configured IPv6-in-IPv4 tunnels can also be used if there is a part of the network that cannot (for whatever reason) be natively connected to the rest of the IPv6 topology. Since these kinds of tunnels have to be configured by hand this only makes sense if there is a requirement for just a few of those tunnels. There is no point in connecting a lot of different isolated hosts or subnets by a manually configured tunnel. There are other tunnelling methods specifically designed for this purpose, such as tunnel brokers, 6to4, Teredo or ISATAP (see below for more details on those). In sites that support IEEE 802.1q VLAN Layer 2 mechanisms, the VLANs can be used to carry IPv6 traffic between disparate subnets in the site, as described in [Cho04b].

Note that IPv6-in-IPv4 tunnels may be used for OSPFv3 routing but not with IS-IS as IS-IS is based on Layer 2 while IPv6-in-IPv4 tunnels are completely Layer 3. For IS-IS to work over a tunnelled link generic encapsulation (GRE tunnels [RFC2784]) can be used.