[walrus01]

I think he means doing NAT on his PC itself (such as is the default network configuration on virtualbox), where it doesn't ordinarily matter what private IP range the VMs are located in. VMs get DHCP leases and a default gateway from the host. The host presents only one IP address and MAC to what it's physically plugged into.

[Godel_unicode]

I know what NAT is, and even if I didn't it was obvious from the post, my question was about the range chosen for the clients. Given that the RFC covering 100.64../10 is specifically for carrier-grade NAT (that's what cgn in my post refers to) I was surprised at the choice to use it instead of one of the other private IP ranges specified in rfc1918, such as 192.168../16.

One place where cgn addressing can trip people up is with DNS; lots of DNS servers (especially the flimsy ones used in lab-in-a-box setups) end up filtering host records for those ranges which can screw up SSH by making the reverse lookup fail, for instance.

Edit: from the text of the RFC -

"""Because CGN service requires non-overlapping address space on each side of the home NAT and CGN, entities using Shared Address Space for purposes other than for CGN service, as described in this document, are likely to experience problems implementing or connecting to CGN service at such time as they exhaust their supply of public IPv4 addresses."""

[walrus01]

I agree that it would be rather unusual to intentionally choose the 100.64 cgnat ranges for guest VMs running on a workstation.