Administrator Guide

1398 IPv6 Routing
On the Dell Networking N1500, N2000, N3000, and N4000 Series switches,
IPv6 coexists with IPv4. As with IPv4, IPv6 routing can be enabled on
loopback and VLAN interfaces. Each L3 routing interface can be used for
IPv4, IPv6, or both. IP protocols running over L3 (for example, UDP and
TCP) are common to both IPv4 and IPv6.
How Does IPv6 Compare with IPv4?
There are many conceptual similarities between IPv4 and IPv6 network
operation. Addresses still have a network prefix portion (network) and a
device interface specific portion (host). While the length of the network
portion is still variable, most users have standardized on using a network
prefix length of 64 bits. This leaves 64 bits for the interface specific portion,
called an Interface ID in IPv6. Depending upon the underlying link
addressing, the Interface ID can be automatically computed from the link
(e.g., MAC address). Such an automatically computed Interface ID is called
an EUI-64 identifier, which is the interface MAC address with ff:fe inserted in
the middle.
IPv6 packets on the network are of an entirely different format than
traditional IPv4 packets and are also encapsulated in a different EtherType
(86DD rather than 0800 which is used with IPv4). The details for
encapsulating IPv6 in Ethernet frames are described in RFC2462.
Unlike IPv4, IPv6 does not have broadcasts. There are two types of IPv6
addresses — unicast and multicast. Unicast addresses allow direct one-to-one
communication between two hosts, whereas multicast addresses allow one-to-
many communication. Multicast addresses are used as destinations only.
Unicast addresses will have 00 through fe in the most significant octets and
multicast addresses will have ff in the most significant octets.
How Are IPv6 Interfaces Configured?
The basic IPv6 protocol specifies two classes of PDU options, both of which
are supported: hop-by-hop options and destination. Although new options
may be defined in the future, the following are currently supported: routing
(for source routing), fragment, router alert, and pad. IPv6 jumbograms (RFC
2675) are not supported. In IPv6, only source nodes fragment. ICMPv6
support of path MTU discovery is therefore supported. IPv6 forwarded or
routed packets are never fragmented by the switch. IPv6 flow labels are
ignored.