Data Communications Networking Devices - 4th Ed.pdf

232 __________________________________________________________ WIDE AREA NETWORKSFigure 2.38 IPv6 provider-based unicast address format. The IPv6 provider-basedunicast address is expected to be the ®rst group of IPv6 addresses to be allocatedThe Registry ID ®eld identi®es the Internet registry responsible for assigningthe address, such as the InterNIC in North America, RIPE in Europe, andAPNIC in Asia. The Provider ID ®eld identi®es the Internet Service Provider.This ®eld is 16 bits in length and is followed by a reserved ®eld of eight bits thatcould be used as an extension to the service provider. The 24-bit Subscriber IDidenti®es a unique user and the assignment of this portion of the provider-basedaddress is the responsibility of the Internet Service Provider. Similar to theProvider ID ®eld, the Subscriber ID ®eld is followed by a reserved ®eld eight bitsin length that will initially be set to zero. This reserved ®eld may be used as anextension to the Subscriber ID ®eld. The remaining 64 bits in the provider-basedaddress will be used in a manner similar to IPv4 Class A through C addresses inthat they will identify a network and host. In actuality, since the ISP is identi®edby the Provider ID ®eld, the Intrasubscriber ®eld can be used by a networkadministrator who is a customer of the ISP to subdivide their assigned block ofaddresses into subnet and station-ID ®elds. The subnet portion would be used toidentify different networks operated by the organization, while the Station ID ®eldwould identify stations on each network.One of the more interesting aspects of IPv6 is the manner by which privatenetworks will eventually be able to be connected to the Internet without requiringmassive con®guration changes as under IPv4. For example, under RFC 1918 theIETF reserved three blocks of addresses for networks that are not connected to theInternet. Although the intention of RFC 1918 is quite admirable, when anorganization decides it's time to connect to the Internet they must obtain anappropriate group of addresses and then recon®gure IP addresses for each networkdevice. Under IPv6 this address reassignment problem is handled far moregracefully due to the ability of allocate link-local use and site-local use addresses.Here the term link references a communications facility such as a frame relay orATM network, a point-to-point leased line or a connection to an Ethernet, Token-Ring, or FDDI network. Thus, a link address represents an isolated device ortransmission facility that has no router connection and is not currently connectedto the Internet, while a link-local address provides a mechanism for connecting thenetwork or device to the Internet without requiring a new address and an addressrecon®guration. This capability is illustrated in the top portion of Figure 2.39which illustrates the general format of a link-local address. Here the uniqueaddress ®eld for connecting the link could represent a LAN MAC address. Sincethe pre®x is 10 bits in length, the remainder of the address is 118 bits in length.Then, if the link-local address is used to connect an Ethernet network, the MACaddress would be 48 bits in length. This would result in a link-local address havingthe 80-bit pre®x FE80 :: followed by the 48-bit Ethernet MAC address. Thus, theuse of a link-local address enables Ethernet, Token-Ring, and FDDI networks tobe connected to the Internet without requiring the recon®guration of networkaddresses.