A. IPv4 addresses are pretty easy. Typically, there’s one IP address for each adapter, which might be an Internet-addressable IP address (unlikely), or a company or home network, and address translation is performed for Internet communications. The IP address could be statically configured, dynamically assigned via DHCP, or automatically generated by the computer if DHCP is unavailable.

Each IPv6 adapter can have a number of IP addresses types:

  • Global Unicast Addresses–These addresses are similar to public IPv4 addresses, routed across the entire IPv6 Internet and allocated by the Internet Assigned Numbers Authority (IANA, at www.iana.org). Global Unicast addresses always have the first three bits set to 001. The following 45 bits make up a Global Routing Prefix, which is unique for each organization. The organization uses the last 16 bits for subnets.
  • Link-Local Addresses–These addresses are the equivalent of Automatic Private IP Addressing (APIPA) for IPv4, which uses the 169.254.0.0/16 network. They’re assigned to hosts that don’t have IP addresses and can't contact a stateful configuration server (such as a DHCP server). Link-local addresses may only be used to communicate with same-network nodes. Link-local addresses all start with fe80; the remaining network address bits are zeroed out because link-local addresses don't use subnets.
  • Unique Local Address–These addresses are the replacement for site-local addresses (which were part of earlier IPv6 standards). They’re designed to be used only within an organization. The first eight bits are always 11111101, meaning all unique-local addresses start with “fd." The next 40 bits make up the global ID, which can be used to identify buildings or locations within an organization. The last 16 network ID bits comprise the subnet ID, allowing multiple subnets within a single location. Make sure the global ID is made with random numbers to future-proof your network in case of a possible merger with another network. If both organizations use "10" as a global ID, you'd have a problem. It's unlikely that two organizations' IDs would overlap if the global ID were made of random numbers.