How ISPs will assign IPs /subnets for end users in IPv6
Consider two scenarios:
The end user has a simple computer, dual-stack
The end user has a LAN behind a dual-stack router
How will ISPs differentiate each one of these kind of users in IPv6?
An user in scenario (1) would receive a single IP, whereas an user in scenario (2) would receive a /64 subnet (not sure if I'm correct on this one).
Will ISPs create 2 kinds of services (one for single IPs and one for subnets)? How does stateless auto configuration play a role here?
Solution 1:
This depends on the class of service being offered. One gets assigned via stateless autoconfig, DHCPv6, or straight up static assignment, the other gets assigned a /64 or bigger. There isn't much difference here. In the v4 world it'd be one IP address and a /29 for the two classes of service, which the ISP is already doing.
The likes of Comcast, residential IP supplier for a large hunk of the US, is planning on passing out a /64 subnet to each subscriber rather than the single IP address they're handing out now.
IPv6 doesn't change anything here. Statically/dynamically assigned single IP addresses will continue to be served from the ISP's netblock, routable blocks will continue to be passed to customer equipment for further routing. It's just that the addresses are bigger.
Solution 2:
There potentially multiple networks here. First the network between the ISP and the user. There the addresses are provisioned with SLAAC (Stateless Address AutoConfiguration) and/or DHCPv6. This address (or these addresses, it's very normal with IPv6 for a device to have multiple addresses) is used by the device connected directly to the ISP (directly connected PC or a router). If a router is used it will request a block of IPv6 addresses for use on the LANs behind the router using DHCPv6-PD (Prefix Delegation). Those addresses can then be used by the router to assign addresses to LANs so that devices connected there can get their own IPv6 addresses.