IPv6Network - IPv6 over Ethernet Ethernet has a long history, and all aspects of the protocol except one have changed somewhere along the way: the cables are now Unshielded Twisted Pair (UTP) or fiber rather than coaxial cable; the speed can now also be 100, 1000, or 10,000 megabits per second in addition to the original 10 Mbps; we now use switches rather than a single shared cable or hubs; and so on.
Putting an IP packet in an Ethernet frame is extremely straightforward: the IPv4 or IPv6 packet simply occupies the “user data” portion of the frame, and the Type field is set to 0x800 (IPv4), 0x806 (IPv4 ARP), or 0x86dd (IPv6) (RFCs 894, 826, and 2464, respectively). Because the user data has a minimum length of 46 bytes, it’s possible that the IP(v4) packet doesn’t occupy the entire Ethernet frame and there are some extra bytes to fill up the minimum length. Because the IP header carries its own length field, this doesn’t lead to problems. If you’ve ever spent a lot of time learning about networks, you may be familiar with IEEE 802.3, SAP, SNAP, and the like. These are further developments of the Ethernet II standard. However, they are completely irrelevant to IPv4 and IPv6, as IP is always encapsulated in Ethernet II frames, as shown in Figure 8-3. This is sometimes called “ARPA encapsulation.”
