Not sure about the large number of lists here... anyway... On Wed, 16 May 2001, Gert Doering wrote:
* IAB/IESG recommended /48. * Use a /128 where it is absolutely known that one and only one device is required, e.g. dialup [<--!!!!!!!] * Use a /64 when sure net will not be subnetted, e.g. a mobile phone given 802.11, bluetooth, etc. ---
I find this thinking, or at least the examples very flawed.
Anyone want to start implementing NATv6 for people whose ISP refuses to give enough addresses to you can't (sub)network your home?
The wording is perfectly clear: if you have more than one device, the ISP MUST give you a /64 (under that policy). If you have more than one subnet, the ISP MUST give you a /48.
I suggest that in this world many ISP's will charge more for the /64 than the /128, and yet more for the /48 than the /64. This means those looking to save money (e.g. students, non-profit organizations, or just cheap people like myself) will probably get the /128, and the next thing you know there's NATng. UltraCheapFastISP User Agreement: "The USER agrees that only a single device will be connected and send data via the PROVIDER's facilities." Personally I think this is fine, but then again I love NAT. :) Is it just me, or does the whole discussion seem academic? Because when many ISP's see the gazillions of addresses they get they're going to allocate it in whatever way they want, and won't bother the RIR's for more space for decades. The only other alternative I see is proactive policing by the RIR's, and I think very few people want that. (I surely don't.) -- Shane "speaking only for myself" Kerr