Illustrated TCP/IP by Matthew G. Naugle Wiley Computer Publishing, John Wiley & Sons, Inc. ISBN: 0471196568   Pub Date: 11/01/98

Chapter 153
Local-Use IPv6 Addressing

This addressing is used exactly as the name implies: locally. This can be subnet local or subscriber local, which gives the two names link local and site local. The slide shows the format of these two addressing types. Notice that both addressing types use the reserved prefix format of FE.

This is an indicator, if you will, such that the internet will not attempt to route packets that are designated as local. You can think of these addresses as the private addresses used in IPv4.

Stations that are not configured using a provider based address or a site local address use the link local address. This is an address that can be used between two stations on a single link or network. This type of address will not be processed by a router, so it cannot span subnets. It can be used by a station that is starting up and does not know its location on the network. Basically, it is the concatenation of its 48-bit MAC address and the well-known link local prefix of FE80::48-bit MAC address. When the make address cannot be used, a serial number or some other unique identification of the card can be used.

A site local address is used to allow a site to configure its network without being connected to the Internet. Unlike IPv4, a site can devise and implement a complete addressed Internet network. This will allow that site to communicate with all interfaces at the site (it may span globally); however, none of these stations may communicate over the Internet. There may be many reasons for this; for example, some companies may not want connection to the Internet until a specified time in the future. I witnessed this at a bank that set up its complete internet based on private addressing IPv4, that is). Problem is it did not use the RFC 1918 private address space allocation to accomplish this. The site was up and operational for two years without a hitch (well, not too many!). When connection to the Internet was desired, the bank had a choice of providing for Network Address Translation (NAT, RFC 1631) or readdressing its network. A lot of thought went into it and based on many factors—scalability, peer-to-peer communication, and others—the bank readdressed its site.

Local-Use IPv6 Addressing

This would not have occurred with IPv6 site local addressing. The addressing allows for any entity to pick any number out of the blue and configure its site (in this case, all of its company locations). If, at some later time, the bank is assigned a global provider address prefix, its network will not have to be completely renumbered.

Site local addresses may not be routed over the Internet, without having a different prefix assigned, such as a global-provider-based prefix. The subnet ID is what you are suspecting. It is an ID assigned to a subnet.